Just add multiple S/PDIF outputs...

"It's worth pointing out that there are potentially serious jitter problems with audio over HDMI which Meridian (for one) has pointed out."

That is the automatic position of every company that is behind the 8 ball with no HDMI solution on the market.

With good design jitter in the carrier is not an issue. Jitter is only important at the input to a DAC or DSP. If the data is clean there anything that happened earlier on is not relevant.

Shawn

It is moot because DD5.1 and DTS bitstream audio is available from the source via toslink or Coax S/PDIF.

It's not moot, because the difference between 12:1 compression (DD) and 3:1 compression (DTS) and the lossless audio formats is like night and day. On a high end system, it is like comparing SD with HD. Digital Dolby, in particular, sounds muddy and harsh compared to a well engineered lossless soundtrack.

The Gefen HDMI matrix and splitter products do not have the bandwidth to support the advanced audio capabilities of HDMI 1.3, or even HDMI 1.1..... but of course, if one sticks to Digital Dolby or DTS, then everything is just fine. (Found out through bitter experience).

So the question is, will the two HDMI outputs of this new video processor PROPERLY transport the uncompressed PCM 5.1 audio??

Until we have properly conforming HDMI 1.3 inputs / outputs, and dual HDMI 1.3 outputs on the video processors, I fear we have to continue our work arounds.

Given that decoding in the player is preferred for enhanced "user experience" (e.g. mixing in commentaries on to the main audio which can only be done on the decoded PCM), transporting the new codec bitstreams will not be required by many. And HDMI v1.1 is perfectly adequate for the transport of multi-channel hi-resolution lossless PCM.

Shawn:

I don't want to veer this thread into a discussion on audio jitter. I agree that jitter is critical at the input to the DAC. I also believe that certain high-end audio manufacturers do a lot better than most of the mainstream consumer electronics companies.

Meridian has stated publically that it does not wish to provide an HDMI upgrade for 1.1 and then another for 1.3. This I accept, but 1.3 hardware is with us now and the clock is ticking. Bob Stuart has also stated that the spec. for HDMI metadata has not yet been locked down. This I do not accept as Meridian is very able to provide firmware updates. So, yes, I think there is an element of stalling.

I don't claim to be an expert on jitter, but the first post by ultima_gtr in the following thread on the Meridian forum is very interesting. You can take it or leave it...

Thread (http://www.meridianunplugged.com/ubb/ultimatebb.php?ubb=get_topic;f=7;t=001044)

"It's not moot, because the difference between 12:1 compression (DD) and 3:1 compression (DTS) and the lossless audio formats is like night and day."

Where did I say a single thing about sound quality differences?

Read my whole post, not pieces parts... you are backing up my point.

You do not gain the ability to hear lossless formats over S/PDIF if the Radiance takes in HDMI audio and spits it out over S/PDIF because S/PDIF can't transport 6 channels of high resolution LPCM audio. That was my whole point. The only thing the Radiance can really take in over HDMI and spit out over S/PDIF are the formats that were already available on S/PDIF anyway.

Shawn

Jim - if possible I would very much like the ability to copy the active HDMI input direct to the second HDMI output without any modification at all. With the best will in the world, VP manufacturers are not high-end audio experts (and vice versa). The Radiance will be my HDMI switching hub for per-input calibration reasons, but I would like my audio processor to be able to get at the original HDMI sources as it is likely to do a better job at audio clock recovery.

This is a feature that the hardware is capable of, and we are considering. It will not likely be in the first production release, but I have had enough requests that it has our attention. Please don't take this as a commitment, but as making it a good possibility this feature can be supported in a future release.

We are getting tremendous response to our RadianceXD and it will form the basis for a number of products going forward. So, we plan to continue to enhance the Radiance feature set for at least several years - just as we have done for our Vision line even three years after introduction. I think we will be adding this feature, I just don't want to commit until we know for sure.

A number of people have asked me to post something on HDMI 1.3. This is my attempt to clear up some of the mystery. Hopefully I don't end up with thousands of posts to respond to on this topic, but here goes:

==============

Lumagen has had a number of questions regarding HDMI 1.3 and if it will be supported in the RadianceXD. Almost all of these inquiries are about if we will support the new HDMI 1.3 audio formats with the RadianceXD. Only a few have been wondering about HDMI 1.3 video formats, but it has come up on occasion. I hope the following helps clear up some of the confusion concerning HDMI 1.3, and what it does, and does not, bring to the party verses the now common HDMI 1.1 used by the RadianceXD. And I’m sure this will stir up a bunch of controversy, but I think it is important to discuss this even knowing others may disagree or have other information I’m not aware of. At least it should reduce the speculation going on about HDMI 1.3.

The RadianceXD, using HDMI 1.1, can support up to 8 channels of un-compressed LPCM audio at 96 KHz, along with bitstream DD/DTS. Even though it cannot support bitstream DD+, bitstream DD-THD or the DTS HD sources at the bitstream level, this seems a rather moot point because, if the HD-DVD or Blu-ray is mastered in "Advanced Content" mode (I think all are) decoding is required in the player. So, they will output the audio as LPCM at up to 96 KHz, which the RadianceXD supports, and not bitstream format. Of course, someone correct me if I’m wrong here, but this is what I understand.

So, the real question is not "Does the RadianceXD support HDMI 1.3 audio?" but "Does the RadianceXD support the uncompressed audio (DD+, DD-THD, and DTS HD) for HD-DVD and Blu-ray, and other sources?" The answer (as best we can determine) is YES. In other words, there is no great advantage in supporting HDMI 1.3, when considering audio with these source formats.

We need to also consider HDMI 1.3 video capability. I believe HDMI 1.3 video is not a factor for current sources, including Blu-ray and HD-DVD, given the HDMI chips that will become available shortly.

The HDMI 1.3 chips I know of becoming available over the next year, or so, support only 12-bits pixel-depth maximum. HDMI 1.1 already supports 12-bit pixel-depth video in the 4:2:2 format. The new HDMI 1.3 format is 4:4:4 (i.e. full-resolution Chroma transmission), which might seem better. However, it's not a significant factor, especially for consumer level sources such as DVD, HD-DVD and Blu-ray as discussed below.

To maximize compression for the minimum loss of quality, consumer video compression schemes use an 8-bit 4:2:0 video format. This means Chroma is half the resolution of luminance in both the horizontal and vertical direction. Before being output from the player this must be up-sampled vertically (for Chroma only) to a 4:2:2 format. The relevant question is, should the player then up-sample to a 4:4:4 format, or should this step be done later in the video pipeline? If all devices in the pipeline have the same performance, it should not matter where this step occurs. Given the less than stellar up-scaling seen in many players, I believe that this should be done late in the video pipeline – either in the video processor, or the display/projector. Said another way, I believe that for interconnect between devices there is no harm, and potentially visible benefit, using the 12-bit 4:2:2 format that is already available in HDMI 1.1, even if the HDMI 1.3 with its 12-bit 4:4:4 format where available.

So, what about 16-bit pixel depth HDMI 1.3 will eventually offer? Sources my someday improve on the 8-bit, overly compressed nature of video sources we are stuck with now. This will probably not be the case for cable or satellite where the money is in quantity, not quality. However, I would be surprised if any consumer sources reached a 12-bit equivalent signal-to-noise ratio. At the other end of the pipeline, displays will not likely have more than a signal-to-noise ratio equivalent of 12-bits either - independent of what is actually claimed by manufacturers. Where more pixel depth is useful is in the actual calculations used in the pipeline. Each stage of calculation adds some level of noise to the image. The video processing chips that are available today max out at 10-bit. The RadianceXD is actually designed to work with higher-precision video-deinterlacing chips when they become available. In the RadianceXD, this chip sits on a small daughter-card that can be replaced with a higher performing chip in the future. So, it is capable of being upgraded where the extra precision is really needed – in the calculations – by replacing this daughter-card and updating the RadianceXD software software.

While I believe HDMI 1.3 has the potential to be an incremental step in the audio/video chain, in my opinion, it is not a significant factor for the next year or two at least. For audio, the new formats do not appear to have a practical use in consumer applications at this point. For video, the existing 12-bit 4:2:2 format should be indistinguishable from the new HDMI 1.3 formats. And you don’t need to go out and buy the even more expensive HDMI 1.3 rated cables, players, amplifiers or projectors just for their HDMI 1.3 capability. One note: While the RadianceXD can pass uncompressed LPCM format, make sure your player and amplifier support LPCM if you want to listen to uncompressed audio. This is a good reason to upgrade your equipment.

Jim,

Thanks for the info. In this thread or an earlier one I think you had mentioned the HDMI was modular and that it could be replaced at a later date much like the Gennum can be replaced. Is that still the case or did that part of the design become fixed?

Thanks,

Shawn

Clear explanation. I agree that people only see an higher number and want that withouth knowing what it means.

Also are the new things in HDMI 1.3 not almost all optional, and not mandatory when you use 1.3 ?
So if it is labeled 1.3, it is not always deep color, bitstream audio and video/audio sychronisation, but perhaps only the wider bandwidth ?

Thanks for the clarification, Jim.
if the HD-DVD or Blu-ray is mastered in "Advanced Content" mode (I think all are) decoding is required in the player.
Decoding in the player is the default, but the player is allowed to offer an option to output bitstream even for "Advanced Content" mode discs. This was explicitly confirmed by Amir a few days ago in the Industry Insiders Thread.

That said, I see no real reason why anyone should need to use bitstream. So IMHO in terms of audio support HDMI 1.3 is not needed.

I believe that for interconnect between devices there is no harm, and potentially visible benefit, using the 12-bit 4:2:2 format that is already available in HDMI 1.1, even if the HDMI 1.3 with its 12-bit 4:4:4 format where available.
If you talk about the connection between source devices and the video processor then I fully agree with you. However, it's quite different when talking about the connection between the video processor and the display! The Gennum chip internally calculates in 10bit YCbCr 4:4:4. And HDMI 1.1 cannot transport that. So you have to downconvert the video data in any case, if you're using HDMI 1.1. While with HDMI 1.3 you could simply pass the final video stream through to the display. Furthermore some displays look better when being fed RGB. And I think 8bit RGB (which is the only RGB mode supported by HDMI 1.1) is somewhat prone to banding artifacts, especially when using video levels. Just imagine the display doing some more processing on the meager 16-235 RGB data...

Hi Jim,

Can the new Lumagen support the lip sync feature of HDMI 1.3? I thought it might be useful if the Lumagen could read the time delay introduced by a source device and add this delay to its own internal delay.

I agree with jacovn. Just because a product has HDMI 1.3 doesn't mean its fully implemented. At this point it seems more like a "catch phrase".

Jim,

Great news that the Radiance architecture will support HDMI pass through. :)

Thanks for the clarification on HDMI 1.3. Not much I hadn't gleaned from elsewhere, but it will be nice to reference your post when required!

Finally, can you confirm that the Radiance XD will output 10-bit 4:2:2 as I believe my JVC HD1 processing is 10-bit and the Gennum chip inside should upsample 4:2:2 competently.

Joel

Finally, can you confirm that the Radiance XD will output 10-bit 4:2:2 as I believe my JVC HD1 processing is 10-bit and the Gennum chip inside should upsample 4:2:2 competently.
Isn't this the subject of a bit of controversy right now? ;) (Yes, I'm bitter that mine is delayed...)

I believe that what Jim is saying is that 10-bit 4:2:2 is in their plans. The question is really going to be whether the JVC knows what to do with a 10-bit signal. Let's hope it does!

Bill

I believe that what Jim is saying is that 10-bit 4:2:2 is in their plans. Just to clarify, I was querying 10-bit 4:2:2 as opposed to 12-bit 4:2:2.

The question is really going to be whether the JVC knows what to do with a 10-bit signal. Let's hope it does!Erm, yes... I guess it's not the end of the world if I have to send it an 8-bit 4:4:4 signal.

Just to clarify, I was querying 10-bit 4:2:2 as opposed to 12-bit 4:2:2.

I think it's safe to assume that it will push 10-bit 4:2:2 as well. The bigger question I have is whether there is an effective way to convert 10-bit 4:4:4 effectively into 12-bit 4:2:2 while restricted to a 10-bit data pipe (does magic happen?).

Can we get some more info on RadianceXS, particularly price? (Sorry if this is too off-topic for a thread about the RadianceXD)

I've been wanting a VP lately and this one looks like it'll be awesome.

Jim,

Thanks for the info. In this thread or an earlier one I think you had mentioned the HDMI was modular and that it could be replaced at a later date much like the Gennum can be replaced. Is that still the case or did that part of the design become fixed?

There are three circuit boards in the RadianceXD that could each be replaced. These are the video input/output board, the audio/misc board and the Video-deinterlacer board.
It is possible to independently change any of these.

We plan to offer an uprade independantly for the video input/output board (to add HDMI 1.3). We don't have a timetable or price set yet. Also, as I posted I don't think there is a great reason to do the upgrade, but we will offer it.

Also are the new things in HDMI 1.3 not almost all optional, and not mandatory when you use 1.3 ?
So if it is labeled 1.3, it is not always deep color, bitstream audio and video/audio sychronisation, but perhaps only the wider bandwidth ?

Yes, the various features of HDMI 1.3 are optional.

What I understand is that the units calling out HDMI 1.3 support the uncompressed bitstream formats and not the deeper video pixel depths. Some may support the deeper pixel depths I just don't know of any at this time.

Thanks for the clarification, Jim.

Decoding in the player is the default, but the player is allowed to offer an option to output bitstream even for "Advanced Content" mode discs. This was explicitly confirmed by Amir a few days ago in the Industry Insiders Thread.

That said, I see no real reason why anyone should need to use bitstream. So IMHO in terms of audio support HDMI 1.3 is not needed.

Thanks for the clarification. Still, I agree with you that the player can decode preventing the need for HDMI 1.3 with no loss of audio quality.

If you talk about the connection between source devices and the video processor then I fully agree with you. However, it's quite different when talking about the connection between the video processor and the display! The Gennum chip internally calculates in 10bit YCbCr 4:4:4. And HDMI 1.1 cannot transport that. So you have to downconvert the video data in any case, if you're using HDMI 1.1. While with HDMI 1.3 you could simply pass the final video stream through to the display. Furthermore some displays look better when being fed RGB. And I think 8bit RGB (which is the only RGB mode supported by HDMI 1.1) is somewhat prone to banding artifacts, especially when using video levels. Just imagine the display doing some more processing on the meager 16-235 RGB data...

I think we agree the source should feed the Lumagen with 4:2:2 at 12-bit, since this is as close to the source format as possible.

I need to clarify that the Gennum does scaling at 4:4:4, but that it does other processes at 4:2:2. In addition we do some processing at 4:4:4 and some at 4:2:2. For example for Y/C delay we upsample the Chroma and then shift it and then down sample again.

Where we might disagree is on whether this is at all visible. Based on what we have seen, this up/down sample of Chroma does not create visible artifacts. This is because the Chroma only has bandwidth of 1/4 the Luma resolution. This is coupled with the human eyes reduced acuity for Chroma. So, while I think artifacts in Luma, such as ringing, are easily seen, I just don't think you can see this Chroma processing.

I agree that CRT projectors look better being feed RGB, but disagree for digital displays. This comes from the fact that virtually all current displays mess with the color space for both RGB and 4:2:2 YCbCr inputs. This means they have to convert the RGB inputs to YCbCr (digital component) and then back to RGB.

With calibration on a digital display, RGB and component will look the same. In fact since virtually all current displays/projectors convert RGB inputs to YCbCr so they can add their infamous red and green push, sending the display 4:2:2 should, if anything, be better than RGB as it has one less conversion --- and you get the 12-bit depth. I should note that I used to recommend RGB until all the manufacturers changed there ways and started messing with the RGB input color space.

So, I have to say I disagree; 4:2:2 is better than RGB for current digital displays. I still believe not having a 12-bit RGB (or 4:4:4) format is moot.

Hi Jim,

Can the new Lumagen support the lip sync feature of HDMI 1.3? I thought it might be useful if the Lumagen could read the time delay introduced by a source device and add this delay to its own internal delay.

You will likely have to wait for our HDMI 1.3 hardware upgrade. We might be able to get it done with the 1.1 chips, but we would have to look closely at it to know for sure, and we won't be looking at this anytime soon.

The time sync feature also presumes the source reports the info and does so correctly. I don't think is a an issue in any case for clean sources like DVDs, HD DVDs and Blu-ray, where audio and video have a pretty consistent (from what I have seen) time sync, as much as it is for satellite and cable. I think it may be some time before this becomes common for either satellite or cable, and even longer before they get it right a majority of the time. Hopefully I'm wrong on this as it would definitely be nice for these sources to have better consistency of delay between audio and video.

Jim,

... can you confirm that the Radiance XD will output 10-bit 4:2:2 as I believe my JVC HD1 processing is 10-bit and the Gennum chip inside should upsample 4:2:2 competently.

Joel

We will actually output 12-bit 4:2:2. To clarify, the video pipeline is a full 10-bits from input up to the color palette used for gamma and grayscale calibration. However, we will have 12-bits out from this color palette, and then keep 12-bits through the rest of the pipeline and output 12-bit 4:2:2.

At first this may not seem to be beneficial. However, when gamma is corrected by the Lumagen, we need to move the output values with reference to the input. Consider a simplified video with 8 levels where you want to change one of the levels by 1%. To do this you need to have 7-bits in the palette output (1 part in 128) even though this simplified input only starts out with 3-bits (1 in 8), or else you can't represent this small change.

As another example, for our actual video pipeline, we might want to change a level of say 512 to 511.25. This requires 12-bit accuracy (1 in 4096). With 10-bits, the 511.25 would be rounded down to 511. If we started with the linear map we use as default, this would mean the sequence is 510, 511, 511, 513. A small flat spot in the curve. This extra bits prevent the flat spots.

I think it's safe to assume that it will push 10-bit 4:2:2 as well. The bigger question I have is whether there is an effective way to convert 10-bit 4:4:4 effectively into 12-bit 4:2:2 while restricted to a 10-bit data pipe (does magic happen?).

To add to the post I just made:

We use the Gennum for deinterlacing and noise reduction. Unlike other vendors, we do the rest of the video processing in our own Lumagen proprietary logic design using a Field Programmable Gate Array (FPGA). So, we can increase the pipeline depth at various points of the pipeline (such as from the palette to the output) when it helps the picture quality. Unlike the other processors out there we are not limited to what the Gennum (or some other vendors chip) can do for any portion but deinterlacing and noise-reduction.

This also allows us to use our proprietary "No-ring" scaling, and add our own extensive video calibration features.

Can we get some more info on RadianceXS, particularly price? (Sorry if this is too off-topic for a thread about the RadianceXD)

I've been wanting a VP lately and this one looks like it'll be awesome.

I can't get into specifics at this point, except to say we do plan a RadianceXS, with one HDMI output and fewer inputs, while having the same video processing pipeline and picture quality.

We do not yet have a schedule nor have we set a price - even internally at Lumagen. We are focused on getting the RadianceXD out to Beta right now and are not doing more than having a few "what-if?" converstions on the RadianceXS. Sorry I can't be more specific just yet.

What we have decided is that we plan to announce RadianceXS specifics at CEDIA in September.

Thanks Jim for all this information !

I just translate all that in french here (on homecinema-fr.com) (http://www.homecinema-fr.com/forum/viewtopic.php?p=170788632#170788632).

Do you know when the RadianceXD will be in a production and commercial state ?

Richard.

We will actually output 12-bit 4:2:2. To clarify, the video pipeline is a full 10-bits from input up to the color palette used for gamma and grayscale calibration. However, we will have 12-bits out from this color palette, and then keep 12-bits through the rest of the pipeline and output 12-bit 4:2:2.


Will there be any compatibility issues if the projector is DVI only?

Mark

Just realised I only ever put up pics of front of my unit earlier in thread. So here are some of back


Full width
http://www.pbase.com/gmfism/image/78687279/large.jpg

Analogue inputs
http://www.pbase.com/gmfism/image/78687273.jpg

Digital In and Out
http://www.pbase.com/gmfism/image/78687278.jpg


I had the lid off the unit when pics were being taken so that's why edges look strange.

Gordon

Two outputs is a great idea and since downscaling is a "less lossy" procedure than upscaling,
lossy is the loss of information. upscaling is not a lossy operation. it adds information.
downscaling removes information and is highly lossy.

its obviously illustrated:

take a 480p image and upscale to 1080p. the image will be of roughly equal quality to the source.

now take a hd-dvd 1080P and downsample to 480p. it will look HEAPS worse than the original.

Hi Jim,

"We plan to offer an uprade independantly for the video input/output board (to add HDMI 1.3). We don't have a timetable or price set yet. Also, as I posted I don't think there is a great reason to do the upgrade, but we will offer it."

I agree that I don't think there is a great reason to upgrade. I just wanted to know if it were possible if say a great reason came down the line later on. Glad to hear it is possible.

Thanks,

Shawn

Nice pics Gordon. Just LOOK at all dem HDMI inputs!

Can you post dimensions?

lossy is the loss of information. upscaling is not a lossy operation. it adds information.

Upscaling adds it's own artefacts and changes the appearance of the image and could therefore be considered "lossy" compared to the fidelity of the original.

Mark

Nice pics Gordon. Just LOOK at all dem HDMI inputs!

Can you post dimensions?

I wonder how Lumagen decided on that number of ports... and I wonder who will be the first to require more! :D

Mark

We will actually output 12-bit 4:2:2. To clarify, the video pipeline is a full 10-bits from input up to the color palette used for gamma and grayscale calibration. However, we will have 12-bits out from this color palette, and then keep 12-bits through the rest of the pipeline and output 12-bit 4:2:2.

At first this may not seem to be beneficial. However, when gamma is corrected by the Lumagen, we need to move the output values with reference to the input. Consider a simplified video with 8 levels where you want to change one of the levels by 1%. To do this you need to have 7-bits in the palette output (1 part in 128) even though this simplified input only starts out with 3-bits (1 in 8), or else you can't represent this small change.

As another example, for our actual video pipeline, we might want to change a level of say 512 to 511.25. This requires 12-bit accuracy (1 in 4096). With 10-bits, the 511.25 would be rounded down to 511. If we started with the linear map we use as default, this would mean the sequence is 510, 511, 511, 513. A small flat spot in the curve. This extra bits prevent the flat spots.Thanks Jim. Very clear (at the second reading :D ). I have my doubts that the JVC HD1/RS1 will accept 10-bit 4:2:2 anyway (yet alone 12-bit 4:2:2), so I guess I'll be outputting 8-bit 4:4:4.

I wonder how Lumagen decided on that number of ports...Basically cost driven I think. Jim sort of explained here (http://www.avsforum.com/avs-vb/showthread.php?p=9132581&&#post9132581) .

We will actually output 12-bit 4:2:2. To clarify, the video pipeline is a full 10-bits from input up to the color palette used for gamma and grayscale calibration. However, we will have 12-bits out from this color palette, and then keep 12-bits through the rest of the pipeline and output 12-bit 4:2:2.Jim, thinking about this again, what happens with 8-bit inputs which are grayscale processed and output at 8-bits? Rounding errors and much less granularity in the calculations?

Upscaling adds it's own artefacts

o rly? like what?

o rly?

Heh, a game player? :D

Depends on how pedantic you want to be about what an artefact is, but upscaling, and particularly non integer upscaling (doubling/quadrupling) cannot preserve sharp edges and thus softens the image - some scalers would add false sharpening to get around this, introducing ringing into the image. It can also exacerbate aliasing by stretching diagonals, making staircasing on diagonals more visible. All such degradation can be considered artefacts.

Cheers,

Mark

Nice pics Gordon. Just LOOK at all dem HDMI inputs!

Can you post dimensions?

Joel, Jim had previously advise me that the dimensions were identical to those of the now discontinued VisionProHD.

HTH

Heh, a game player? :D

Depends on how pedantic you want to be about what an artefact is, but upscaling, and particularly non integer upscaling (doubling/quadrupling) cannot preserve sharp edges and thus softens the image - some scalers would add false sharpening to get around this, introducing ringing into the image. It can also exacerbate aliasing by stretching diagonals, making staircasing on diagonals more visible. All such degradation can be considered artefacts.

Cheers,

Mark

not entirely true. 480 to 1080 is a large enough leap to preserve all original content. you can go back and forwards all day and the image would barely change. if one uses nearest neighbour scaling then how can you say sharpness is lost: you just have 6 times more of the same original pixel ;)

probably a bit too OT sorry, but i didnt like the inference that in the real world upscaling is lossy.

Holy #$^&,

Got enough HDMI inputs there?

Actually, I've got 6 HDMI sources at this point in time, so there's something to it now...

not entirely true. 480 to 1080 is a large enough leap to preserve all original content. you can go back and forwards all day and the image would barely change. if one uses nearest neighbour scaling then how can you say sharpness is lost: you just have 6 times more of the same original pixel ;)

probably a bit too OT sorry, but i didnt like the inference that in the real world upscaling is lossy.

Whether you like it or not is irrelevant ;) Nobody is disputing that the original data is preserved. What is relevant is whether or not the additional image information added by the scaling can introduce visible "losses" in fidelity; ringing, softness and exacerbated aliasing are all examples of such losses. And trying to keep tenuously OT, it's one of the reasons that Lumagen are proud of their scaling, because it introduces no ringing; which implies that other scaling algorithms do. Now whether anybody can see these side effects is also an interesting point, and certainly with 1080 sources, compared to 480 sources, it can be more difficult, but they are there all the same.

Cheers,

Mark

Joel, Jim had previously advise me that the dimensions were identical to those of the now discontinued VisionProHD.Thanks Joel, Joel ;)

not entirely true. 480 to 1080 is a large enough leap to preserve all original content. you can go back and forwards all day and the image would barely change. if one uses nearest neighbour scaling then how can you say sharpness is lost: you just have 6 times more of the same original pixel ;).
NN upsampling is of no use with images in motion pictures as it creates severe aliasing.

I think we agree the source should feed the Lumagen with 4:2:2 at 12-bit, since this is as close to the source format as possible.
Definitely.

Where we might disagree is on whether this is at all visible. Based on what we have seen, this up/down sample of Chroma does not create visible artifacts. This is because the Chroma only has bandwidth of 1/4 the Luma resolution. This is coupled with the human eyes reduced acuity for Chroma. So, while I think artifacts in Luma, such as ringing, are easily seen, I just don't think you can see this Chroma processing.
Ok, fair point. I haven't tested that myself, but I think I trust your opinion there. That means: I can accept HDMI 1.1 to be good enough for YCbCr output.

But would you agree that HDMI 1.3 is a worthwhile upgrade for people who have to use RGB (video processor -> display) for whatever reason? Provided that the display has HDMI 1.3, too, of course, and makes actually use of the added bits.

NN upsampling is of no use with images in motion pictures as it creates severe aliasing.

This is important ! I understand the temporal aliasing.

I am implying that the video source(480i) is NOT the canon we base the benchmark upon. The original is the film. a 480i print has artifacted the original film, and by using 1080p upscaling we can produce an image closer to the original film than the 480i is.

My arguement is that 1080p upsampled is a truer replica of the film source than the 480i will be. So the claim that it is a lossy procedure is misrepresented.

Thanks Jim for all this information !

I just translate all that in french here (on homecinema-fr.com) (http://www.homecinema-fr.com/forum/viewtopic.php?p=170788632#170788632).

Do you know when the RadianceXD will be in a production and commercial state ?

Richard.

Thanks for the translation effort.

As to schedule:

First let me say that we will be shipping the first Beta units this week. We just took an additional two week hit getting production parts and then we just happened to take our parts to the assembly house the very day they sent ALL there engineeres to a week long class. Cost us a week since I had made a small change and they couldn't start the run until the engineers were available again. Fun times in manufacturing for sure.

We now have the first production run of circuit boards in house, so we will be able to ship the first Beta units out this week. This months shipment is limited in size, but we will have a larger build in June and another in July.

I should again let people know that our Beta units are production hardware and preproduction software.

Now to you specific quesiton: We plan production release mid-September, and will offically announce the RadianceXD at CEDIA.

Will there be any compatibility issues if the projector is DVI only?

Mark

RadianceXD will support DVI displays.

This brings up another point that some amplifiers get wrong:

If the display is DVI, we will still report that we are HDMI to the source. So if you have a DVI display we can send DVI out to that display on one output, and HDMI on the second output to either an HDMI amplifer or an HDMI display.

I really like this feautre since one doesn't need to buy a new display to be able to use HDMI audio.

I wonder how Lumagen decided on that number of ports... and I wonder who will be the first to require more! :D

Mark

Good question.

Originally I planned on four HDMI inputs, 2 HD component, 2 SD/ED component, 2 SVideo and 2 composite.

I went from four to six HDMI when Josh Lehman convinced me a high-end scaler needed six HDMI. I noticed that Ofer is already glad we did since he has 6 HDMI sources.

We had a number of current Lumagen owners with CRT projectors request more analog video inputs, so I increased the number to four of each component, SVideo and composite. I should also note that one component input can be combined with two of the composite inputs to provide a single RGBHV input.

Came out to video inputs 18 inputs.

I figured we needed as many audio inputs as we had video, so we have 18 audio (including HDMI) inputs as well.

I also got a lot of requests for a LVTTL IR-command input port and trigger outputs. So we added the LVTTL IR port and two 12-volt trigger outputs.

Can you post dimensions?

The unit is 17" wide by 3.75" high by 9.75" deep.

For rack mounting, when the feet are removed, the unit changes to 3.5" high for a 2U rack height.

Jim, thinking about this again, what happens with 8-bit inputs which are grayscale processed and output at 8-bits? Rounding errors and much less granularity in the calculations?

For the 8-bit in and 8-bit out case, we bring the data in, process it at 10-bits, send it though the gamma table with 12-bit out, and then we would "dither" it to 8-bits before output.

Dither is a way to reduce the visual effect of a reduced bit-depth. It works quite well and is the same as what we do now in the 10-bit to 8-bit conversion in our Vision series.

... But would you agree that HDMI 1.3 is a worthwhile upgrade for people who have to use RGB (video processor -> display) for whatever reason? Provided that the display has HDMI 1.3, too, of course, and makes actually use of the added bits.

I don't see how.

For a CRT with an RGB analog connection, we will do an external DAC and will use HDMI 4:2:2 to the DAC box to get the extra bit depth.

For a DVI display, there is no option more than 8-bits of RGB.

For an HDMI display, you have 12-bit 4:2:2 available already and I hold to my earlier post on why this is a great way to go and as good or better than RGB.

I think I need you to give me an example of why you would ever want to do this, because I can't think of one.

This is important ! I understand the temporal aliasing.

Spatial aliasing here, not temporal.

I am implying that the video source(480i) is NOT the canon we base the benchmark upon. The original is the film. a 480i print has artifacted the original film, and by using 1080p upscaling we can produce an image closer to the original film than the 480i is.
My arguement is that 1080p upsampled is a truer replica of the film source than the 480i will be. So the claim that it is a lossy procedure is misrepresented.
Lossy or not is not the point. You can make the upsampling mathematically lossless. More interesting is if it looks better or worse upsampled. New artifacts visible? Old artifacts gone or less visible? Looking more analogue? More digital? Sharper? Softer? Smoother? Rougher? Etc.

HDMI 1.3 in the video processor means you can pass the advanced sound streams in their original format for decoding on your prepro/receiver, if its also 1.3. The main drawback I see to using HDMI 1.1 on a video processor is that doing so compels you to send your HDMI to the receiver/prepro FIRST, and this is almost always a problem. It might mess with the video data, it might have only a single output etc. As an end user you are discouraged from sending it to the Lumagen first, because you want the advanced codecs handled by the prepro/receiver and NOT by the source device.

While 1.1 can send LPCM to the receiver, some people prefer to have the prepro/receiver do ALL THE processing. Look how many HDMI 1.1 devices can't handle more than 5.1 LPCM streams for example.

Spatial aliasing here, not temporal.

NN filters destroy you temporally not spatially. Its the dramatic change per frame that renders it useless (its similar to horizonal lines on interlaced film appearing and disappearing. I wont comment more because its the wrong thread :o

to be OT does the VXP do 1080P down-conversion to 768P? Also does Gennum publish the scaling method used?

NN filters destroy you temporally not spatially.mhafner is correct - the problem nearest-neighbor scaling introduces is a spatial one. There may be secondary effects caused because of the spatial problems introduced, but the scaling process itself is purely spatial and so introduces only spatial artifacts. Now granted, the drawbacks to this type of scaling can become more visible with motion, but it's the spatial scaling artifacts that are being accentuated. They're there whether there's motion or not, and the right program material makes this obvious.

Also does Gennum publish the scaling method used?Seeing as the RadianceXD is using Lumagen-proprietary scaling, not that provided in the Gennum chip (which is mentioned clearly and repeatedly in this thread), it's rather moot, no? Perhaps Jim might want to say more about Lumagen's scaling, though.

- Dale Adams

NN looks like crap even without the temporal dimension.

They're there whether there's motion or not, and the right program material makes this obvious.
- Dale Adams

i think you will find an 1080p NN scaled image imperceptably different to a 480P image. you cant test it though because you dont have a 480p display (maybe crt monitor).
NN is infact the most accurate scaling method when taken in integer ratios. but i will re-iterrate that accurate is not necessarily optimal (remember the source in movies is not the 480p dvd. the source is the master, and then before that, the real scene).

For an HDMI display, you have 12-bit 4:2:2 available already and I hold to my earlier post on why this is a great way to go and as good or better than RGB.
In most cases: Yes. But I wouldn't bet on that there'll never be the odd display where things are different. Well, whatever... :)

i think you will find an 1080p NN scaled image imperceptably different to a 480P image. you cant test it though because you dont have a 480p display (maybe crt monitor).
NN is infact the most accurate scaling method when taken in integer ratios..
Hm. It violates the assumptions for sampling and Nyquist. A correct analogue version of the digital picture (and upsampled versions approaching it) does not have the pixel perfect edges anymore of say text where NN seems to be the only algorithm preserving detail. But it's false detail. The picture is band limited and that limit must show when upsampling (edges become ultimately ramps etc.).

i think you will find an 1080p NN scaled image imperceptably different to a 480P image. you cant test it though because you dont have a 480p display (maybe crt monitor).You think wrong on both counts - demonstrably so. Look, this is way off topic and is potentially derailing the discussion from the RadianceXD. If you want to start a discussion on NN scaling or other scaling techniques and their fidelity (or lack thereof), why don't you start a dedicated thread for that? This is, after all, exactly the right forum for this type of discussion.

- Dale Adams

Just realised I only ever put up pics of front of my unit earlier in thread. So here are some of back


Full width
http://www.pbase.com/gmfism/image/78687279/large.jpg

Analogue inputs
http://www.pbase.com/gmfism/image/78687273.jpg

Digital In and Out
http://www.pbase.com/gmfism/image/78687278.jpg


I had the lid off the unit when pics were being taken so that's why edges look strange.

Gordon

I didn't realize that they're not using BNC anymore ala the Lumagen HDQ. I like the BNC connector since it engages and disengages without putting a lot of stress on the actual unit. Some RCA connectors clamp on so well that I feel it may just pull the unit's connector right off. :(

You think wrong on both counts - demonstrably so. Look, this is way off topic and is potentially derailing the discussion from the RadianceXD. If you want to start a discussion on NN scaling or other scaling techniques and their fidelity (or lack thereof), why don't you start a dedicated thread for that? This is, after all, exactly the right forum for this type of discussion.

- Dale Adams

i dont think im wrong, what im saying is paradependant of nyquist. but ill stop posting on this thread before heads roll. :p

For a CRT with an RGB analog connection, we will do an external DAC and will use HDMI 4:2:2 to the DAC box to get the extra bit depth.

Jim I will need one of these. Will they ship with the unit or be available later?

Martin

Martin,

They will be available at a later date. If you require one now then there are a few available. SIM2, Keene, Spatz, Dtronics and I think Gefen all do them.

Gordon

Call me ill informed, but wouldn't an external DAC that converted HDCP encrypted data into an analog form violate the HDCP license?

Bill

I would also love a version of the backpanel that has BNCs rather than RCAs.

Bill it would but Lumagen will make sure that the DAC will not do that. It will only work on non HDCP material. I plan to use HD-SDI from source to VP to avoid HDCP so it won't be a problem.

Martin

Bill it would but Lumagen will make sure that the DAC will not do that. It will only work on non HDCP material. I plan to use HD-SDI from source to VP to avoid HDCP so it won't be a problem.

Martin


Unfortunately Radience does not have HD-SDI inputs, how are you olanning to run your HD-SDI source to VP?

I didn't realize that they're not using BNC anymore ala the Lumagen HDQ. I like the BNC connector since it engages and disengages without putting a lot of stress on the actual unit. Some RCA connectors clamp on so well that I feel it may just pull the unit's connector right off. :(

We are considering a Pro version with BNCs. It would end up with fewer analog inputs since the BNCs are much larger.

We have had perhaps an 85% vote for RCAs, and about 15% vote for BNCs. Personally I like BNC connectors.

BTW: Under normal use the RCA recepticles on the Lumagen will not pull out. I can't speak for the plugs.

Key Digital Blaster (http://www.digitalconnection.com/products/video/kdsdi1080p.asp)

I really wished the Radience would have 2 HD-SDI and onboard DAC but alas. I now need 3 extra boxes (2 HD-SDI + DAC) and have precious little room to put them.

Martin

The Gefen HDMI matrix and splitter products do not have the bandwidth to support the advanced audio capabilities of HDMI 1.3, or even HDMI 1.1..... but of course, if one sticks to Digital Dolby or DTS, then everything is just fine. (Found out through bitter experience).Could you please elaborate on the problems you experienced?

I'm using several Gefen switchers in my too-complex system, and though I have do have some complaints (switchers respond to stray IR from unrelated components; no proper support for 480i over HDMI even with updated firmware), I haven't noticed any issue with (for example) the LPCM output of the Toshiba HD-XA1 HD-DVD player when switched through a Gefen device to my Denon HDMI-capable A/V receiver.

Seeing as the RadianceXD is using Lumagen-proprietary scaling, not that provided in the Gennum chip (which is mentioned clearly and repeatedly in this thread), it's rather moot, no? Perhaps Jim might want to say more about Lumagen's scaling, though.

- Dale Adams

I have tried to let others work on this sub-thread, but think I should chime in at this point.

We do not use the Gennum scaling, as Dale points out. We have looked at it, and while it matches other non-Lumagen scalers for quality, we believe ours is dramatically superior.

I have to agree with mhafner, Dale, and others. NN is not a good scaling algorithm for video. It violates Niquist sampling and will introduce really bad artifacts. About the only good thing that could be said is if one used NN to scale up to an exact power-of-two integer multiple of the input and then back down you could reproduce the original image. However, that is not what we are trying to do.

Scaling must produce some additional artifacts. Our job is to keep these in the "not visible" category. Pretty much everyone but Lumagen uses traditional 2D FIR filters. These cause ringing. I think ringing is the worst scaling artifact and so we designed our own that does not generate ringing.

However, every scaling step will cause some softening of the image as well. NN does keep a sharp edge, but at an overwhelming negative cost of false edges. I could not bring myself to watch an image scaled from 480p to 1080p done with NN for both the spacial and temporal aliasing it creates.

To overcome the softening required to filter/scale the image, all scalers try to re-sharpen the image post-scaling. We do this as well. However, we avoid ringing while doing so. Others actually enhance the ringing (either on purpose or by mistake) to try for a "Sharp" image. This can make the image appear sharper on first glance, but when you examine the image you find it is hash and looks less realistic, and less three dimensional.

Balancing the filter to minimize softening in the filter, eliminate ringing, and then sharpen the image post scaling is a very fine tight-rope to walk. From what I hear, people like what we came up with.

Key Digital Blaster (http://www.digitalconnection.com/products/video/kdsdi1080p.asp)

I really wished the Radience would have 2 HD-SDI and onboard DAC but alas. I now need 3 extra boxes (2 HD-SDI + DAC) and have precious little room to put them.

Martin

Martin:

Are you able to just convert 1080I HD-SDI to 1080i DVI without any scaling done by the Digital Blaster?

Are you able to just convert 1080I HD-SDI to 1080i DVI without any scaling done by the Digital Blaster?
AFAIK yes. The Digital Blaster can definitely do it for 480i and 576i. And it does support HD-SDI and 1080i. So I think direct passthrough should work for 1080i, too.

Key Digital Blaster (http://www.digitalconnection.com/products/video/kdsdi1080p.asp)

I really wished the Radience would have 2 HD-SDI and onboard DAC but alas. I now need 3 extra boxes (2 HD-SDI + DAC) and have precious little room to put them.

Martin

Yes I agree, I would even be happy with just 1 HD-SDI input.............

Glad I'm not alone.

Martin

Balancing the filter to minimize softening in the filter, eliminate ringing, and then sharpen the image post scaling is a very fine tight-rope to walk. From what I hear, people like what we came up with.

It strengthens the theory that scaling is an art, not a science. It also means no scaler will suit everyone.

Perhaps other companies dont deal with it to the level you guys have.

AFAIK yes. The Digital Blaster can definitely do it for 480i and 576i. And it does support HD-SDI and 1080i. So I think direct passthrough should work for 1080i, too.
Doesn't HD-SDI require two BNCs? Again, I'm probably missing something on this. The box may work on the signal, but it does not look like it actually has the necessary inputs. Is the second BNC for HD-SDI a sync or loop of some sort?
(Never really looked into HD-SDI...)

http://www.computermodules.com/broadcast-systems/hd-sdi-format-converter.shtml
http://www.gefen.com/images/HDSDItoDVIBack1.jpg
vs.
http://www.keydigital.com/images/Video%20Processors/Digital_Blaster_v1_h.jpg

Bill

Hd-SDI is one cable one BNC connector

Hd-SDI is one cable one BNC connector
As I said, never really looked too seriously into it. Most of the stuff that has breezed by had the two BNCs.

Jim:

Could you provide any further information on the outboard DACS ?

In the older HD-SDI standard, you would need dual link BNCs to carry 1080P whereas the new standard can do 1080P with a single BNC.

For more details:

http://en.wikipedia.org/wiki/HD_SDI

Jim: Could you provide any further information on the outboard DACS ?


I haven't completed the design yet, so this is in flux. However, plans include:

- HDMI input

- It will not function for HDCP encrypted sources. Sorry, our HDCP license prevents this.

- It can be set to output any one of component, RGBHV, or RGsB.

- At this point I will be using the same 10-bit DAC we use in the Vision series.

- DAC will always run at 148.5 MHz like the Vision series. This provide 2X over-sampling at 720p/1080i, and 1X sampling at 1080p.

- BNC ouput connectors

So the DAC is the same quality as in the Vision series, but we will be feeding it a higher quality signal. So it should look better than the Vision series analog out.

Thanks Jim

Is component output from Blu-ray or HD-DVD HDCP encrypted or just the HDMI output?

Just HDMI.

Shawn

No reason not to use an even better DAC. Just raises the cost a bit. Whoever needs the DAC will buy it regardless of whether it cost $40 more than one with a less good DAC. The problem is most will need a DAC box or card that strips HDCP. A US manufacturer just can't do it because it would violate the license and subject the violator to hugh financial penalties. However, stripper DACs are more or less readily available from Korean sources. Of course Marquee guys are waiting for the latest plug in card version. Should be available very soon.

No reason not to use an even better DAC. Just raises the cost a bit. Whover needs the DAC will buy it regardless of whether its cost $40 more. The problem is most will need a DAC box or card that strips HDCP. a US manufacturer just can't do it because of the license and the financial penalties. However, stripper DACs are more or less readily available from Korean sources. Of course Marquee guys are waiting for the latest plug in card version. Should be available very soon.


Yes Mark, looking for the Best quality DAC, Aside from HDCP issue, wonder how Lumagens will compare to the latest plug in cards from Moome and John?

What's the latest on the new Moome card for the Marquee? Last I heard he was revising the hardware.

Don't know, he has a full functioning input card for the G90

"I believe that what Jim is saying is that 10-bit 4:2:2 is in their plans. The question is really going to be whether the JVC knows what to do with a 10-bit signal. Let's hope it does!"

According to tstites it does not.

Shawn

"I believe that what Jim is saying is that 10-bit 4:2:2 is in their plans. The question is really going to be whether the JVC knows what to do with a 10-bit signal. Let's hope it does!"

According to tstites it does not.Shawn, I've been following the RS1 threads since their inception and, although he means well, the amount of incorrect information posted by Tom Stites leaves him with very low credibility IMO.

As it happens, I suspect that he will be right on this, but I've asked my dealer to contact Akira Kobayashi - head of the RS1 design team - to get a definitive answer. Akira (Alex) is a very nice guy BTW.

It's worth pointing out that there are potentially serious jitter problems with audio over HDMI which Meridian (for one) has pointed out. I'm sure that other (high-end) audio manufacturers are also concerned about this. HDMI simply wasn't designed as a low jitter carrier. It might not be a big deal with vanilla DD and DTS bitstreams, but I'd be concerned if PCM was extracted. Consequently I will continue to use the S/PDIF outputs from my various sources straight into my G68 processor, until Meridian (eventually) comes out with its HDMI solution for encrypted audio.

Jim - if possible I would very much like the ability to copy the active HDMI input direct to the second HDMI output without any modification at all. With the best will in the world, VP manufacturers are not high-end audio experts (and vice versa). The Radiance will be my HDMI switching hub for per-input calibration reasons, but I would like my audio processor to be able to get at the original HDMI sources as it is likely to do a better job at audio clock recovery.

What about the fact that the G68 (I have one too) is limited to 85 m/s of audio delay, which is simply not enough to cope with the amount of processing time required for HD and some SD material - that is certainly the case with the HDQ anyway? I still don't have an HD DVD player, and will probably sit on the fence for as long as Meridian do. If I was to pick up a Radiance when it is available for sale what would be the best way of hooking up my current DVD player (Arcam DV-88) which has no audio delay facilities. I know it is a little long in the tooth now, but it does seem a little foolish to buy a more capable DVD player when something far better is just around the corner. The HDQ may not be the best SD processor in the world but it certainly improves things on my TH50-PHD8.

Tim

"As it happens, I suspect that he will be right on this, but I've asked my dealer to contact Akira Kobayashi - head of the RS1 design team - to get a definitive answer. Akira (Alex) is a very nice guy BTW."

If you get an answer please let me know.

Thanks,

Shawn

"As it happens, I suspect that he will be right on this, but I've asked my dealer to contact Akira Kobayashi - head of the RS1 design team - to get a definitive answer. Akira (Alex) is a very nice guy BTW."

If you get an answer please let me know.

Thanks,

Shawn
Also ask him if he has heard of Papst fans, and whether they could do a running change... ;)

Tim,

Jim has posted elsewhere, possibly on this thread, that the Radiance processing is MUCH faster than in Vision series. Your 85ms will cover it

Gordon

Thanks Gordon,

That will save having to re-wire all the audio when the time eventually comes. I have enough spare inputs on the Meridian to be able to setup multiple sources with different delays. BTW I haven't completely ignored your advice about Meridian DVD players, but I'd still like to take a look at one when the time comes as the audio section will be fantastic.

Tim

The RadianceXD is really two-tone. Silver (again really light gray) in themiddle with black end-caps and case.

There will be a two-color LED. Red will indicate off. When popwer is on, we will have the default as a green "on" indicator (default on the Vision series is no-light except on command receipt).

Jim,

I know I'm not the only person who will ask this, but will you ever offer a black front panel? ALL my other equipment is black, so the XD would stand out like a sore thumb. Should I wish to get it professionally painted, how easy is it to remove?

Tim

Edit: I've just seen that you have no plans to offer a black front panel at the moment, so I guess I'm just looking for an answer to my second question

Tim,

I'm going to see how simple it is to have one modified soon....

Gordon

So I'm not the only person that's asked then?

Tim,

Jim has posted elsewhere, possibly on this thread, that the Radiance processing is MUCH faster than in Vision series. Your 85ms will cover it

Gordon

FWIW, with a Meridian 861 and a Lumagen ProHDP I am using the following lipsync delays:

Toshiba HD-DVD: 8ms
Sony Blu-Ray: 6ms
Meridian 800 DVD: 72ms(!)

Mark

yes one other person has asked me. I think that once folk see it in the flesh it wont be such an issue. Mine sits amongst other black pieces of equipment and it doesn't stand out.

Gordon

I know what you mean Gordon but I do have 10 monolithic black slabs of equipment and when the HDQ gets swapped out it would be nice to replace it with another black slab. I'm currently looking for a 1U 2 drive enclosure to connect to my Sky HD box. Silver will look OK but black will look so much better.

Tim

Will the XD have a audio delay control? If so how much (and how much is needed for scaling to 1080P)?

Will the XD have a audio delay control? If so how much (and how much is needed for scaling to 1080P)?

Per the first post in this thread it has audio delay and genlocked the delay is 1.25 frames.

Shawn

Genlocked?

Thanks, Shawn.

From the Lumagen FAQ:

Q: What is genlock and how do I use it?
A: Genlock is the ability to exactly lock the output rate to the input's rate or to a multiple of the input rate. It was added to the software with the 072005 update so if you're running an older sw revision you don't yet have the feature. Press "menu 0903" to display your sw revision onscreen. For a typical setup you might be running an NTSC source (which is 59.94 hz nominally) into the Lumagen and outputting a 59.94 hz rate to your display. The clocks of both the NTSC source and the Lumagen devices are never going to match exactly, one will be to some degree faster than the other. If they're very close you may get a dropped (or doubled) frame once every 25 minutes and you may never notice it depending on what you're watching. In some cases they may differ enough you might get a dropped/doubled frame every couple of minutes and chances are you'll eventually see that. With genlock on the clock for the output is very slightly varied in order to keep the rates locked and avoid any dropped/doubled frames. Depending on your setup it may or may not be noticeable as you can see from the above description.

In cases where you're inputting a 50hz source and displaying it at a 59.94hz rate genlock can not be achieved. You would see a "disabled" status for genlock (your video would also be fairly juddery for such a mismatch as well). To check the status press the "ok" button on our remote several times when no onscreen menu is present. Other possible states are OFF, LOCKING and LOCKED.

Genlock is able to lock onto multiples of 24 (actually 24/1.001) or 25 for NTSC or PAL sources so you can also run your output at 24Sf, 47.95hz, or 71.93 hz for NTSC inputs and achieve lock or 75 hz output for a PAL source. You'd normally only want to run at these 24/25 multiples if the source is a true film source such as a DVD movie.

Another benefit of using genlock is it does slightly reduce the video delay (aka lipsync delay) going through the video processor for NTSC/PAL sources by about 1/2 a field.

Genlock can be enabled/disabled on a per-input memory basis under the IN->CONFIG->CNTRL->GENLCK menu.

Will the Radiance have the ability to "shift" a 2.35 image within a 16:9 frame (easily) ? If not yet, please do consider adding this feature. This is VERY useful since u can have a nice masking for both the 16:9 and 2.35. And not to mention to bring down the projected image to a comfortable level. (I'd imagine achieving this via the remote: hitting 2.35 for the input aspect ratio, and a "mem d" for the shift. Or even "Mem X" be associated to 2.35 aspect input automatically. )

Thank you.

Thanks guys!

Will the Radiance have the ability to "shift" a 2.35 image within a 16:9 frame (easily) ? If not yet, please do consider adding this feature. This is VERY useful since u can have a nice masking for both the 16:9 and 2.35. And not to mention to bring down the projected image to a comfortable level. (I'd imagine achieving this via the remote: hitting 2.35 for the input aspect ratio, and a "mem d" for the shift. Or even "Mem X" be associated to 2.35 aspect input automatically. )I've got a feeling that this can already be done with the Vision range, but I might be wrong. I would like this feature too.

"I've got a feeling that this can already be done with the Vision range, but I might be wrong. I would like this feature too."


Yes, I did something like this. It is not a one touch feature though.

I first used the Lumagen with a JVC G10 which was a 4x3 projector. I used it in a constant height setup with an ISCOII. The Isco effectively turned it into a 16:9 projector and the lens was always in place. For 1.78 material I fed the projector its 1365x1024 native resolution but had output AR set to 1.78 because of the lens.


To move from 1.78 to 2.35 material I used the power zoom in the projector along with another output memory set for 1365x768 (output AR 2.35)which the projector displayed as is. As part of this I needed to move the active picture in the 2.35 picture to the top of the screen (edge aligned projector) so that the zoom would then zoom down/out to fill the 2.35 screen. With the custom output timings and adjust the various parameters I was able to move the picture to the top of the panel. This was several years ago so I forget exactly which parameters this needed but I think it was the front/back porch settings. After I got this right I saved the configuration and assigned that output to one of the input memories and just selected that when watching 2.35.

Shawn

No with vision line there is no easy way to do it. I think Shawn's case is special because he needs to power zoom his PJ (also you need to adjust the front/back porch and since it's zoomed it can reach the range). As of now u can also move it by a certain combination of TOPL/BTMR/ZOOM but it's not only not a 1/2 button case but also conceptually not really correct.

For this feature the DVDO approach seems ok. When u hit input AR of 2.35, and your current output AR is 16:9, you then have the capability to move up/down. In addition, when input AR is 4:3 (and output is 16:9), you can even move left/right. The logical reason that u can "move" is because in AR != display AR. It shouldn't be considered a case where I need to "re-center" the input scan area.

"think Shawn's case is special because he needs to power zoom his PJ (also you need to adjust the front/back porch and since it's zoomed it can reach the range)."

The zoom had nothing to do with making the picture edge aligned, the Lumagen did that. I needed it edge aligned (as the projector was edge aligned to the screen) so that when zoomed the picture expanded to fill the screen. If I hadn't edge aligned the picture when I zoomed the picture would just move down and not fill the top of the screen.

So it can be done, just not a one two three thing to setup. Once it is setup and saved to an independent output memory it is easy to recall. But it is not as simple as just moving it around like the DVDO apparently does.

Shawn

Will the Radiance have the ability to "shift" a 2.35 image within a 16:9 frame (easily) ? If not yet, please do consider adding this feature. This is VERY useful since u can have a nice masking for both the 16:9 and 2.35. And not to mention to bring down the projected image to a comfortable level. (I'd imagine achieving this via the remote: hitting 2.35 for the input aspect ratio, and a "mem d" for the shift. Or even "Mem X" be associated to 2.35 aspect input automatically. )

Thank you.

Image shift is a planned feature for the Radiance and it's in the new menu as a to be done item.

Best regards,
Randy Freeman

Jim,

I know I'm not the only person who will ask this, but will you ever offer a black front panel? ALL my other equipment is black, so the XD would stand out like a sore thumb. Should I wish to get it professionally painted, how easy is it to remove?

We might reconsider the all-black front panel if there is a lot of interest. We tried to have the best of both worlds with our two-color front panel, but I understand it is not perfect for all. I personally prefer black as well.

If we continue to get votes for this, it is straight forward to have the annodize on the non-black area stripped and black annodization added.

I will have to see if my local silk-screen company can handle a curved surface. I can also check with our laser-etch folks to see if they can do a curved surface. If either of these pans out, it will be reasonable to offer black.

I would probably have to charge extra to cover the extra steps.

So, keep those votes for all-black coming in.

Will the XD have a audio delay control? If so how much (and how much is needed for scaling to 1080P)?

Beta will not have the Audio delay built in. That will be in a release around production time frame, give or take.

With genlock on, the delay is about 1.25 field times. Add up to one frame time for the non-genlocked case. In the US this is 20 mS or 36 mS, much less than the Vision series, and dramatically less than some competing technologies.

I think an attractive front panel (e.g., the two color front panel) is important as the default case.

I'm happy to hear there might be two options.

Beta will not have the Audio delay built in. That will be in a release around production time frame, give or take.

With genlock on, the delay is about 1.25 field times. Add up to one frame time for the non-genlocked case. In the US this is 20 mS or 36 mS, much less than the Vision series, and dramatically less than some competing technologies.
Am I correct in thinking that genlock will work if you are sending the audio over HDMI or dig coax to the radiance and then out to the AVR.
Does audio have to come in over the HDMI to work with genlock

Am I correct in thinking that genlock will work if you are sending the audio over HDMI or dig coax to the radiance and then out to the AVR.
Does audio have to come in over the HDMI to work with genlockGiven Genlock involves varying the internal clock rate of the VP to ensure no periodically dropped frames, I hate to think what the potential consequences are for audio clock recovery by a surround processor.

All the more reason for allowing the selected input to be copied without modification to the second output. Are the two output clocks independent? Getting a little worried here!

Beta will not have the Audio delay built in. That will be in a release around production time frame, give or take.

With genlock on, the delay is about 1.25 field times. Add up to one frame time for the non-genlocked case. In the US this is 20 mS or 36 mS, much less than the Vision series, and dramatically less than some competing technologies.

Hmm. For HD-DVD, going via a ProHDP, with audio through a Meridian digital system, I only need 3ms lipsync right now (4ms for Blu-Ray) to get perfect lip sync (I guess the Meridian processing adds it's own delay). Am I going to find that the Radiance is *so* fast that the audio will always be behind the image? Can we have an image delay option?!

Mark

Am I correct in thinking that genlock will work if you are sending the audio over HDMI or dig coax to the radiance and then out to the AVR.
Does audio have to come in over the HDMI to work with genlock

Genlock has to do with the video only. So, the audio is not affected by genlock settings.

Audio from the HDMI, COAX, optical and analog will all work as well when using (or not using) genlock.

Jim, I want 1920x1080p72 output

Thanks, what I meant was will the audio delay only work if the audio comes in on HDMI and genlock is applied

Given Genlock involves varying the internal clock rate of the VP to ensure no periodically dropped frames, I hate to think what the potential consequences are for audio clock recovery by a surround processor.

All the more reason for allowing the selected input to be copied without modification to the second output. Are the two output clocks independent? Getting a little worried here!

You bring up an interesting point.

Even thought audio flows through the Radiance in a completely independent path from video, it must be encoded by the HDMI output chip. Since this would use the genlocked clock -- which does vary slightly over time -- it is likely the genlock feature would affect audio clock jitter.

To avoid this, you can set the second HDMI output to "audio only" and we will use a stable crystal controlled clock to generate clock for the blank video at 720p used to carry the audio, and so provide the most stable audio clock possible.

This has been the plan, but thinking on your point, I agree with you about allowing the second output to be a direct copy of the HDMI input. I will add this to our to-do list. We would use the video clock and the audio clocks from the HDMI input chip (through buffers) to directly drive the second output HDMI chip.

NOTE: There is only a single audio path through the RadianceXD. If you want a "direct" connection for the second HDMI output audio, the main output audio would have to match. That is, audio is delayed on both HDMI outputs or neither.

Thanks, what I meant was will the audio delay only work if the audio comes in on HDMI and genlock is applied

Audio delay works from any audio source, and wether or not genlock is on.

Do note that if genlock is off the audio delay would still be a fixed delay, but the video delay will vary over time. It would go from about 1.25 to about 2.25 fields slowly over a period of from several minutes to about 20 minutes depending on how the input and output clocks match.

Hmm. For HD-DVD, going via a ProHDP, with audio through a Meridian digital system, I only need 3ms lipsync right now (4ms for Blu-Ray) to get perfect lip sync (I guess the Meridian processing adds it's own delay). Am I going to find that the Radiance is *so* fast that the audio will always be behind the image? Can we have an image delay option?!

Mark

Mark:

Not sure we can delay the video. Might be able to add a field delay or so, but would need to consider if we can do this before commiting to it. This really will have to wait until after production release to even consider this.

If you end up switching out your CRT for a digital display, the display itself will add addition video delay of between 1 and 2 fields typically.

Jim, I want 1920x1080p72 output and are you still thinking about the blending option? that would be great to have! maybe as a special order option i'd pay extra for that!.

Athanasios

Jim, I want 1920x1080p72 output

Our HDMI output chips are rated at 165 MHz. So, this is not impossible for a digital display. For a CRT, assuming the total blank percentage for H and V, the clock needs to be much higher than this 165 MHz max. In addition, we design all out internal paths for 148.5 MHz. So this would be a large undertaking to up the clock rate in our FPGA, and it would steal gates from other functions.

Still, there may be a way. For a digital display that could take very short bank times, we would not be much over 150 MHz. Still a large task, so I'm not sure we would do this.

For CRT users we might be able to add a frame buffer inside the DAC box, and then set the RadianceXD to output 1920x1080@24 for film and 1920x1080@60 for video (or 50 Hertz in PAL countries). Then the DAC frame buffer could repeat frames for film to get to 1920x1080@72, and output the video at 1920x1080@60.

I have not designed the DAC yet. Since it is really for die-hard CRT projector folks with their G90's, and equivalent projectors, this might make sense. It would add significant cost to the DAC to be able to do this though.

So, here's a poll: If you plan to buy an external DAC from Lumagen for your Radiance, please vote from the following two options. If you don't plan to buy a Lumagen external DAC, please don't vote.

Option 1: Straight DAC with output to 1920x1080p@60 Hertz for somewhere around $400.

Option 2: Smart DAC that can output up to 1920x1080@75 Hertz for somewhere around $800.

To avoid hijacking this thread, please email your votes to us at Lumagen support.

Mark:

Not sure we can delay the video. Might be able to add a field delay or so, but would need to consider if we can do this before commiting to it. This really will have to wait until after production release to even consider this.

If you end up switching out your CRT for a digital display, the display itself will add addition video delay of between 1 and 2 fields typically.

Jim, you'll be happy to know that I retired my CRT late last year :D I'm on the Radiance Beta list so should know if this is going to be a problem soon enough...

Mark

... are you still thinking about the blending option? that would be great to have! maybe as a special order option i'd pay extra for that!.

Athanasios

The best I can say is we are still thinking about it. The issue is how many units this feature sells verses the effort to do the design. Also, it's extra gates and so might steal from other more common functions.

I like the idea of this feature, so it is still on our "under consideration" list.

You bring up an interesting point.

Even thought audio flows through the Radiance in a completely independent path from video, it must be encoded by the HDMI output chip. Since this would use the genlocked clock -- which does vary slightly over time -- it is likely the genlock feature would affect audio clock jitter.

To avoid this, you can set the second HDMI output to "audio only" and we will use a stable crystal controlled clock to generate clock for the blank video at 720p used to carry the audio, and so provide the most stable audio clock possible.

This has been the plan, but thinking on your point, I agree with you about allowing the second output to be a direct copy of the HDMI input. I will add this to our to-do list. We would use the video clock and the audio clocks from the HDMI input chip (through buffers) to directly drive the second output HDMI chip.:) I am smiling. A lot! :)

We might reconsider the all-black front panel if there is a lot of interest. We tried to have the best of both worlds with our two-color front panel, but I understand it is not perfect for all. I personally prefer black as well.

If we continue to get votes for this, it is straight forward to have the annodize on the non-black area stripped and black annodization added.

I will have to see if my local silk-screen company can handle a curved surface. I can also check with our laser-etch folks to see if they can do a curved surface. If either of these pans out, it will be reasonable to offer black.

I would probably have to charge extra to cover the extra steps.

So, keep those votes for all-black coming in.


Jim, add my name to the list of users who would greatly prefer black...

Jim, add my name to the list of users who would greatly prefer black...

Me too. Black is better since it tends to match my other components. I am not sure how the current case will fit in.

- Rich

Took longer than we wanted, but I am pleased to note that ...

We shipped the first batch of RadianceXD Beta units today.

For everyone on the Beta list, please be patient until we get yours ready. The demand for Beta has been much higher than we expected.

Thanks to everyone for your support, and great suggestions for improving the Radiance line of processors.

The best I can say is we are still thinking about it. The issue is how many units this feature sells verses the effort to do the design. Also, it's extra gates and so might steal from other more common functions.

I like the idea of this feature, so it is still on our "under consideration" list.

Right now i'm testing a TVone unit that has dual scaling engins and has the blend feature. it is definatly a slick machine but not very user freindly like the lumagen,guess i'm just used to yours(have two!) The price is still high $7200 but less expensive than the other blender most use the Analog Way Di Ventix $13,000 cheapest i found. so even if it is an add on feature at a slight premium over the normal cost i think alot of crt'rs would jump on it! there are many who are stacking now and a blend would give them alot more use of the phosphore and better resolution, brightness ect.

And i also think this would be a nice feature for digital projectors, no more special lenses, just add another projector, alot easier than what it is for us Crt'rs.

Any one else wanna vote for this feature as well? ! :D

http://www.avsforum.com/avs-vb/showthread.php?t=851150

Athanasios

We might reconsider the all-black front panel if there is a lot of interest. We tried to have the best of both worlds with our two-color front panel, but I understand it is not perfect for all. I personally prefer black as well.

If we continue to get votes for this, it is straight forward to have the annodize on the non-black area stripped and black annodization added.

I will have to see if my local silk-screen company can handle a curved surface. I can also check with our laser-etch folks to see if they can do a curved surface. If either of these pans out, it will be reasonable to offer black.

I would probably have to charge extra to cover the extra steps.

So, keep those votes for all-black coming in.


I vote for black!

congratulations Jim!

Will beta users be able to post some preliminary info and impressions here?

Are the Beta's out yet?

If there are any beta users in NYC (or the surrounding areas) that would be interested in comparing the Lumagen to the Crystallio please PM me for a completely discrete test that will not jeapordize your non disclosure with Lumagen. Just you and me, THX

And you work for?

Seriously. Even if you are completely genuine, this would breach NDA unless you have signed one too.

Not much point in comparing beta unit to anything. Wait for production then do your own comparison. As VK says, it would be a violation of the NDA folk have signed.

Gordon

One thing I'm desperately interested in knowing is how good the new Gennum chip's noise reduction algorithms are. I mean the deinterlacing capabilities of the Gennum are well known. But nobody knows for sure yet how good the new noise reduction algorithms are. Will they be able to compete with the Algolith stuff?

"One thing I'm desperately interested in knowing is how good the new Gennum chip's noise reduction algorithms are. "

Ditto. I have not seen much discussion about that and pretty much none for the new chip.

Shawn

The NDA is very specific. It is designed to get fully working and bug free software out there. Lumagen doesn't want a public discussion about product comparisons,discovered bugs and how their product is not yet bug free. The beta program is to gather user experience, bug discovery, bug duplication in the plant, and the development of fixes so that when the beta program is completed a well working product ends up in the hands of the non beta consumer. Also new features could be developed and added during the beta period making any comparisons not especially valid. Lumagen doesn't want to have its non beta customers have to wait a long time for fixes. Better to be adding even more features after non beta introduction rather than making fixes. :)

The Radiance XD looks great but it occurred to me that as time goes on, I need only digital inputs and outputs for video sources. I wonder if it makes sense to offer a lower end Radiance Digital Only?

- Rich

I am genuine and a private comparison between videophiles that is not to be discussed publicly is not a violation of anything. If, for example, I was a beta tester and owned a Crystalio, would I be in violation doing an A/B comparison between two different products in my own home with my own equipment?

Again, this is not a public discussion or a "shootout" it is a video fanatic who is not on the beta list looking to see if what Lumagen is putting out is better than what I already own.

My opinion, my viewing and my business..not for public debate.

"If, for example, I was a beta tester and owned a Crystalio, would I be in violation doing an A/B comparison between two different products in my own home with my own equipment?"

If you demoed it to someone that didn't sign the NDA, yes.

That is what a non disclosure is.........

"Again, this is not a public discussion or a "shootout" it is a video fanatic who is not on the beta list looking to see if what Lumagen is putting out is better than what I already own."

Then wait for it to be released and do a comparison then when the software for the Radiance is in a non-beta state. The whole point of it being in beta is that it isn't complete/ready yet. A comparison against anything in that state really won't tell much of use if you have to wait for a production unit anyway.

Shawn

A dealer is allowed to do a customer hands off demo. For the customer to buy, a NDA is required and Lumagen's approval of the customer must be obtained. In other words, the customer would become in essence another beta tester

I know its a little weird, but this is a beta phase. Beta should end sept/october. During the beta phase, the product is only being made available to beta testers and those testers must sign and are subject a NDA. The product is only available to beta testers.

And beta testers pay MSRP for the product, or what?

Dan,

Per Jim in post 101:

"If anyone is interested in being a RadianceXD Beta test site, please email sales@lumagen.com and we will try to accommodate you. As with the Vision series, since Beta units have production hardware, and can be updated to full production with a software update, we do charge the same price for Beta as production."

Shawn

Dan. It's sort of like you become an insider. While you aren't paid, you get the product months before its production release and help the Lumagen team develop the product. Its a special feeling when as a beta tester, you make a suggestion and see it get implemented.

congratulations Jim!

Will beta users be able to post some preliminary info and impressions here?

Thanks Ofer.

We do have an NDA with each Beta users this go around. So, this is not going to happen for a while. I plan to allow some public comments in about two months after we get more functionality in the unit.

Might change my mind and allow some comments sooner.

The Radiance XD looks great but it occurred to me that as time goes on, I need only digital inputs and outputs for video sources. I wonder if it makes sense to offer a lower end Radiance Digital Only?

- Rich

A digital only input unit has been discussed. We could do this, but the analog section doesn't add a huge amount and given volumes we need for each different product, we are thinking along the lines of having a few analog inputs in a lower end unit.

We are still in the discussion phase for the next version of Radiance, so all suggestions are welcome.

Given the Beta NDA dialog, I thought I should talk a bit about what and why of the NDA.

First, Mark is correct that if you contact a Lumagen dealer who is a Radiance Beta, they can give you a demo - assuming you are interested personally and not for business - such as a competitor. Demos only though, no comparisons for now.

Since Beta users are under NDA they can't talk about any comparisons they make for themselves, but there is nothing to prevent them making them. In fact we want them to, and to let us know. If we fair well as expected, great, if not, we can spend time improving. They just can't share results with others who are not under NDA.

The idea behind the NDA is as several have said. We don't want comparisons until the software is closer to production. When we did the Vision Beta, there were several public comparisons made that did not apply once we hit production. We believe that the fact that the comparison was made on a Beta unit got lost, while the comparisons themselves lived on.

The picture quality on the RadianceXD is already great as we have the Gennum for deinterlacing, and our Vision no-ring scaling, in place already. However, we plan further enhancements and want comparisons to include these. In addition major features have yet to be added, and we think comparisons should include these as well.

Thanks for your patience in waiting a couple more months for public comparisons to start appearing.

Jim: As always thanks for keeping everyone in the loop on info. Would you say Aug or later when we'll be able to buy these bad boys? I realize this is a ballpark ETA. I'm trying to budget ahead for this purchase.

Thanks
Ron

Jim: As always thanks for keeping everyone in the loop on info. Would you say Aug or later when we'll be able to buy these bad boys? I realize this is a ballpark ETA. I'm trying to budget ahead for this purchase.

Thanks
Ron

While we are shipping the first Beta units to distributors, dealers and a few individuals (most of whom were Vision Beta sites as well), we are open to Beta orders from others. We are currently scheduled to release to production in September.

Whether you want production or Beta, I suggest getting your order in as we have a good sized pre-order list for both.

Some times a manufacturer makes a post that is seminal in that it is a clear reflection of the very essence of the company and the essence of its head. No bull, the pure truth and the complete rationale for an action. No spin and in a way no one could disagree.

We as forum members and AV enthusiasts are honored to have JRP participate here and to have someone like him working hard to bring us something many of us need or want in a manner and with a degree of professionalism that we should all strive for in our owm lives. I think the man would make a good judge. I am not really sure I should be posting this but it is what I feel and it is intended to be a compliment for somthing very special in the AV industry.

While we are shipping the first Beta units to distributors, dealers and a few individuals (most of whom were Vision Beta sites as well), we are open to Beta orders from others. We are currently scheduled to release to production in September.

Whether you want production or Beta, I suggest getting your order in as we have a good sized pre-order list for both.

I am currently on a pre-order list with AVS. Thanks for the response.

We are still in the discussion phase for the next version of Radiance, so all suggestions are welcome.
My opinion: I don't need any analog inputs. But having 4 optical audio inputs might be useful.

Just a little idea: How about adding an AC3/DTS decoder and a high quality DAC for multichannel audio? That would allow many many people to keep using their old receivers which don't have HDMI audio support yet. Maybe as an external box? HDMI in (with AC3/DTS/PCM support), 7.1 analog out. Don't know, maybe it's a bad idea...

I am currently on a pre-order list with AVS. Thanks for the response.

I am on the pre order list with AVS also, but not sure if that is beta or production.

Either way, I am sure this is going to be a great unit. I have postponed my RS1 order until the Radiance ships.

People, please make sure to honor the NDA.

Madshi,

I don't know about that. I still have some analog devices here and there (granted, except my SDTV STB, these are getting quite rare...).

I don't know about that. I still have some analog devices here and there (granted, except my SDTV STB, these are getting quite rare...).
I didn't mean to say that the next Radiance shouldn't have any analog inputs. I was really just meaning to say that *I personally* don't need any. But the general trend probably goes to less (not none yet) analog and more digital connections.

I didn't mean to say that the next Radiance shouldn't have any analog inputs. I was really just meaning to say that *I personally* don't need any. But the general trend probably goes to less (not none yet) analog and more digital connections.

I do not need any analog inputs or outputs. That's just me, but it may be an interesting model to provide a lower end product.

- Rich

I do need analog inputs. I'm still playing older game consoles, and am looking forward to the XD's low latency option.

I'm lucky enough to have 4x HDMI sources and an HDMI display. A few of the analogue inputs will be useful from time to time - e.g. S-video from my camcorder and occasional watching of old videos - though the XD will be rather overkill on the analogue inputs for my requirements.

Jim,

The existing Vision units can auto select a memory, and therefor an output memory config, based on the input frame rate to handle PAL/NTSC.

As an expansion to that (Auto Independant-Film ?) could basically the same thing occur for 24p sources? And maybe the same thing for film based sources on 480i or 1080i? For straight 24p this should be pretty straight forward, for 480i/1080i I don't know how well this would work on mixed material.... one wouldn't really want the output memory auto changing often since they will loose the picture as it resyncs with the display.

Thanks,

Shawn

Shawn,

I think for mixed content on interlaced sources that sort of flip-flopping res/refresh change is a bad idea for the reasons you mention. For 24Hz input I think it's a good idea.

Gordon

Jim,

The existing Vision units can auto select a memory, and therefore an output memory config, based on the input frame rate to handle PAL/NTSC.

As an expansion to that (Auto Independant-Film ?) could basically the same thing occur for 24p sources? ...

We have expanded the input resolution/rate "sub-memory" concept. Instead of four (three fixed and one "other"), the Radiance line will have eight submemories. While the "Other" entry will remain a catch-all, seven of these can be reprogrammed. For example, if you wanted a PC calibration for a specific resolution, or perhaps a calibration of 720p24 for Apple HD video. Since there are four memories each with 7 programmable sub-memories, you could have up to 28 input resolutions calibrated for a given input, plus a catch-all "Other."

Since each input has different configuration memories, the Apple PC input could include 720p24, while you leave the defaults for your multi-region HD DVD player.

By default these are:
- 480i/p60
- 576i/p50
- 720p60
- 1080i50
- 1080i60
- 1080p24
- 1080p60
- Other

As you can see there is no more discrete NTSC verses PAL modes. We think this is a more seemless integration of these two different standards as compared to our Vision series. It also gives our customers who do need both twice as many configuration memories to play with, as compared to the Vision series.

For Beta units 480i/p (and 576i/p) are linked in one sub-memory. We plan to allow them to be split in a future release.

I think Analoge inputs are still needed with all the displays stil using anloge, especialy first generation HDTV's(crt) and CRT front projectors. Since even the HD-DVD and Blu-Ray players still have these outputs. With out analoge those who do not have hdcp compliant displays will not be able to use the radianceXD, unless they have access to a HDCp stripper before the VP.

Dont for get to vote for blending option on the new Radiance too ! here is the link !

http://www.avsforum.com/avs-vb/showthread.php?t=851150

Athanasios

Hi Jim,

Just to make sure I understand this.... basically you mean that for any one of this input rates:

- 480i/p60
- 576i/p50
- 720p60
- 1080i50
- 1080i60
- 1080p24
- 1080p60
- Other

Each and every one of them has their own Mem A,B,C,D memory slots? So basically to do what I was suggesting above you could in effect do this with a single memory position.

So for 1080p60 input its mem A could be set to Output 1 (1080p60) while for 1080p24 input its mem A could be set to Output 2 (1080p24)...etc...etc?

If so.... NICE! Just be sure the 'Copy' function works! ;)

Thanks,

Shawn

Jim,

This one is a more out there suggestion but for a later date how about zone based convergence correction?

Something like maybe center and four corners for zones (or more if you are feeling daring) and allow the R/G/B pixels to be shifted left/right up/down. You would probably need some sort of dead area between zones or something but this could be *very* useful on three chippers.

To start out even just whole screen pixel shift for R/G/B (like the RS-1) would help a bunch of Sonys and such.

Shawn

Just a little idea: How about adding an AC3/DTS decoder and a high quality DAC for multichannel audio? That would allow many many people to keep using their old receivers which don't have HDMI audio support yet. Maybe as an external box? HDMI in (with AC3/DTS/PCM support), 7.1 analog out. Don't know, maybe it's a bad idea...

An interesting thought. Video processing circuitry runs at a very high clock rate and therefore generates a reasonable amount of ground noise. It is much better to have your audio DAC in a clean enclosure where the video noise doesn't have a chance to affect the audio signal-to-noise ratio. Since the RadianceXD is targeted at enthusiast and high-end theaters, it makes more sense to me to keep audio separate. In fact I am thinking the lower price point of Radiance will not have audio except for HDMI in and out. Since our delay is short, the ability to delay all the audio source is not so important, and most of our customers have very good amplifiers/receivers to switch and process audio.

In fact, audio is only in the Radiance as a competitive feature. In my opinion, video and audio should always be kept separate. I consider myself an audiophile, and it pains me to see that the only way to get uncompressed multichannel audio is through HDMI. With the extra jitter this incurs it is quite possible that this uncompressed stream sounds worse than current DD and DTS streams through SPDIF. As discussed earlier in this thread, HDMI will likely add jitter to the audio stream. An excellent amp can have circuitry to reduce this jitter to some degree, but this is usually reserved for very high-end equipment. It is better to have a low-jitter clock in the first place.

Since HDMI is the way these non-compressed formats are being supported, HD DVD and Blu-ray player should have two HDMI outputs. One for the video, and one for audio. The audio HDMI output should use the audio clock with blank video at some rate that makes sense using the audio clock. Using the audio clock would allow the jitter to be reduced.

Vendors could even design HDMI parts optimized for audio using the audio clock. If they did, the HDMI connection could have even lower jitter than SPDIF. This is because the audio clock could use the differential clock pair of HDMI. and since the audio is not encoded onto a single wire, this clock would get two benefits: Differential clock and dedicated clock wires.

Just to make sure I understand this....

Each and every one of them has their own Mem A,B,C,D memory slots? So basically to do what I was suggesting above you could in effect do this with a single memory position.

So for 1080p60 input its mem A could be set to Output 1 (1080p60) while for 1080p24 input its mem A could be set to Output 2 (1080p24)...etc...etc?


Yes, you are correct.

Got my test unit setup and running, and man, Jim and Pat really knocked this one out of the park Every spec I wanted is in there, and even with this early beta software, it is looking outstanding.

Good Job!

...how about zone based convergence correction?

Something like maybe center and four corners for zones (or more if you are feeling daring) and allow the R/G/B pixels to be shifted left/right up/down. ...

To start out even just whole screen pixel shift for R/G/B (like the RS-1) would help a bunch of Sonys and such.


A whole pixel shift is possible, and we are considering adding this feature (not a committed feature).

However, I'm not sure we will add any other convergence features. Good idea though. Certainly worth keeping on the radar.

Hi Jim:

I hope that you reconsider having an HD/SD-SDI input, even if its availble as an optional module.

Radiance looks great can't wait for mine

An interesting thought. Video processing circuitry runs at a very high clock rate and therefore generates a reasonable amount of ground noise. It is much better to have your audio DAC in a clean enclosure where the video noise doesn't have a chance to affect the audio signal-to-noise ratio. Since the RadianceXD is targeted at enthusiast and high-end theaters, it makes more sense to me to keep audio separate. In fact I am thinking the lower price point of Radiance will not have audio except for HDMI in and out. Since our delay is short, the ability to delay all the audio source is not so important, and most of our customers have very good amplifiers/receivers to switch and process audio.
At first I thought "noooo" when I read your idea of maybe dropping audio ins. But after thinking about it, maybe you should really drop everything which is not absolutely necessary, so you can go down with the price as much as possible. Of course it hurts if you drop something. But on the other hand people will be glad about every $50 the price is reduced. And people who need all the connections, can then still buy the more expensive Radiance. Maybe having 4-6 HDMI ins and 1 HDMI out would already do the trick? Ouch, Ofer won't like that... :D

BTW, do you have the possibility to artificially "consume" more time for video processing than you really need? I'm asking this because of the following reason: Imagine 2008 comes a new Gennum chip which you'll integrate into the Radiance (thanks to the modular design). This new chip may have a super duper new feature which suddenly produces 200ms audio delay. Of course it's a problem if audio delay is sometimes 200ms and sometimes only 30ms, if audio delay must be handled in the receiver. BUT you could solve this problem nicely if you were able to make sure that the Radiance will *always* produce 200ms delay, even if the super duper new feature is turned off. Hmmm... Am I talking crazy? Basically it would be nice if you could make sure that the audio delay is always constant - even if a future Gennum chip has more varying delays depending on the activated features.

In fact, audio is only in the Radiance as a competitive feature. In my opinion, video and audio should always be kept separate. I consider myself an audiophile, and it pains me to see that the only way to get uncompressed multichannel audio is through HDMI. With the extra jitter this incurs it is quite possible that this uncompressed stream sounds worse than current DD and DTS streams through SPDIF. As discussed earlier in this thread, HDMI will likely add jitter to the audio stream. An excellent amp can have circuitry to reduce this jitter to some degree, but this is usually reserved for very high-end equipment. It is better to have a low-jitter clock in the first place.
Yeah, e.g. Arcam strictly sais they won't add HDMI audio support unless they can manage to somehow "remove" the jitter. And Anthem already does resample the HDMI audio, IIRC. So the middle/high end audio companies seem to agree with you.

Since HDMI is the way these non-compressed formats are being supported, HD DVD and Blu-ray player should have two HDMI outputs. One for the video, and one for audio. The audio HDMI output should use the audio clock with blank video at some rate that makes sense using the audio clock. Using the audio clock would allow the jitter to be reduced.
That would be nice, but I honestly don't expect that we'll see that in any but the most expensive high models... :(

BTW, do you have the possibility to artificially "consume" more time for video processing than you really need? I'm asking this because of the following reason: Imagine 2008 comes a new Gennum chip which you'll integrate into the Radiance (thanks to the modular design). This new chip may have a super duper new feature which suddenly produces 200ms audio delay. Of course it's a problem if audio delay is sometimes 200ms and sometimes only 30ms, if audio delay must be handled in the receiver. BUT you could solve this problem nicely if you were able to make sure that the Radiance will *always* produce 200ms delay, even if the super duper new feature is turned off. ...

We will have the audio delay specified relative to the video delay. So if you set "Audio delay offset" to 0, the actual audio delay with genlock on would be approx 20 mS.

If we did an update with some super new chip with added video delay, you would still set "Audio delay offset" to 0, but we would then delay audio as appropriate to match the new video delay.

An interesting thought. Video processing circuitry runs at a very high clock rate and therefore generates a reasonable amount of ground noise. It is much better to have your audio DAC in a clean enclosure where the video noise doesn't have a chance to affect the audio signal-to-noise ratio. Since the RadianceXD is targeted at enthusiast and high-end theaters, it makes more sense to me to keep audio separate. In fact I am thinking the lower price point of Radiance will not have audio except for HDMI in and out. Since our delay is short, the ability to delay all the audio source is not so important, and most of our customers have very good amplifiers/receivers to switch and process audio.

In fact, audio is only in the Radiance as a competitive feature. In my opinion, video and audio should always be kept separate. I consider myself an audiophile, and it pains me to see that the only way to get uncompressed multichannel audio is through HDMI. With the extra jitter this incurs it is quite possible that this uncompressed stream sounds worse than current DD and DTS streams through SPDIF. As discussed earlier in this thread, HDMI will likely add jitter to the audio stream. An excellent amp can have circuitry to reduce this jitter to some degree, but this is usually reserved for very high-end equipment. It is better to have a low-jitter clock in the first place.

Since HDMI is the way these non-compressed formats are being supported, HD DVD and Blu-ray player should have two HDMI outputs. One for the video, and one for audio. The audio HDMI output should use the audio clock with blank video at some rate that makes sense using the audio clock. Using the audio clock would allow the jitter to be reduced.

Vendors could even design HDMI parts optimized for audio using the audio clock. If they did, the HDMI connection could have even lower jitter than SPDIF. This is because the audio clock could use the differential clock pair of HDMI. and since the audio is not encoded onto a single wire, this clock would get two benefits: Differential clock and dedicated clock wires.

Carry that idea one step further to include an analog volume control, and I can get rid of my receiver altogether. Source-->Radiance-->Lumagen Simplified Pre/Pro-->Amp-->Happiness. :)

Bill

A whole pixel shift is possible, and we are considering adding this feature (not a committed feature).

However, I'm not sure we will add any other convergence features. Good idea though. Certainly worth keeping on the radar.
Add my vote to having a whole pixel shift (horizontal and/or vertical) to correct for panel misalignment. I see no reason to upgrade to the Radiance unless it has this feature (hint, hint). :) The Silicon Optix Geo chip is supposed to have extensive convergence features but I doubt it will show up in an affordable standalone solution anytime soon.

Add my vote to having a whole pixel shift (horizontal and/or vertical) to correct for panel misalignment. I see no reason to upgrade to the Radiance unless it has this feature (hint, hint). :) The Silicon Optix Geo chip is supposed to have extensive convergence features but I doubt it will show up in an affordable standalone solution anytime soon.

Convergence fixing would be a major draw for me.

Me, too..

A pixel shift might make convergence or optical errors less worse but it wouldn't fix them. Any one or more integer pixel shift would have to be global accross the entire pixel grid. Usually on a three chip machine, the divergence of the colored lines (whether caused by panel misconvergence or by lens deficiencies) are not uniform accross the grid. Often a single piel shift would make things better in spots but worsen it elsewhere.

I guess it depends how bad your three-chipper is. If it's bad and linear, then such a feature would be priceless.

However, for those of us with the RS1/HD1 and its pixel shift function, given the eye's (lack of) sensitivity to chroma, I doubt it would be worth it unless it was a non-linear shift.

"Often a single piel shift would make things better in spots but worsen it elsewhere."

Hence the suggestion for making it zone based. 5 zones (center area and each corner) would allow one to better dial in say a bad bottom left without effecting top right which was already fine.

Shawn

I recall using DILARD with my first JVC digital..THe pixel shift was really helpful for the poor panel alignement. My QUalia is off a pixel on red and a 1/2 pixel on blue. The misconvergence is uniform across the screen. So, this would help (in my case).

I think that, realistically, the non-linear warping we're talking about isn't going to happen with the Radiance. IIRC, it's a feature of the Realta but not the Gennum chips.

"Often a single piel shift would make things better in spots but worsen it elsewhere."

Hence the suggestion for making it zone based. 5 zones (center area and each corner) would allow one to better dial in say a bad bottom left without effecting top right which was already fine.Any convergence correction system which does not apply the same correction to the entire image has to be capable of fractional-pixel correction. Otherwise you'd see a very clear demarcation between the different zones. The requirement to shift the image by fractional-pixel increments and by different amounts across the image requires 95% of a full-fledged scaling engine.

Also note that the correction would need to be applied in RGB space. Many (most?) of today's video processors scale in YCbCr space and some don't operate directly on 4:4:4 signals.

The combination of these requirements means that unless the current scaler is working directly on RGB signals, there would need to be the equivalent of a second scaler to perform the convergence correction. I figure it always helps to know just what it is you're asking for. ;)

- Dale Adams

Sounds like the sort of challenge Lumagen just love to sink their teeth into :D

Mark

Thanks Dale. You saved me a long post as to how difficult it would be.

On a related note, shading correction would be nice... :cool:

I tried to read all the 24 pages to look for my answer but my eyes started bleeding..

When is the Radiance coming out on the market to buy?

When is the Radiance coming out on the market to buy?Around September. CEDIA timeframe. You can sign up to the beta program now though.

Hi Trebor,

You can buy a beta unit now if you are prepared to help in the debugging process. You'll need to sign an NDA no doubt and then give feedback to Lumagen. There is no fixed date that the firmware will be called production but Jim has said on is thread that they are working for a timeframe around CEDIA USA as Viruskiller has stated.

It is worth noting that once unit is "production" firmware it is still Lumagens intention t continue to develop the unit and add performance enhacing features.

Gordon

Thanks, I found the info on the beta program.

Video processing circuitry runs at a very high clock rate and therefore generates a reasonable amount of ground noise. It is much better to have your audio DAC in a clean enclosure where the video noise doesn't have a chance to affect the audio signal-to-noise ratio. Since the RadianceXD is targeted at enthusiast and high-end theaters, it makes more sense to me to keep audio separate. In fact I am thinking the lower price point of Radiance will not have audio except for HDMI in and out. Since our delay is short, the ability to delay all the audio source is not so important, and most of our customers have very good amplifiers/receivers to switch and process audio.

In fact, audio is only in the Radiance as a competitive feature. In my opinion, video and audio should always be kept separate. I consider myself an audiophile, and it pains me to see that the only way to get uncompressed multichannel audio is through HDMI. With the extra jitter this incurs it is quite possible that this uncompressed stream sounds worse than current DD and DTS streams through SPDIF. As discussed earlier in this thread, HDMI will likely add jitter to the audio stream. An excellent amp can have circuitry to reduce this jitter to some degree, but this is usually reserved for very high-end equipment. It is better to have a low-jitter clock in the first place.

Since HDMI is the way these non-compressed formats are being supported, HD DVD and Blu-ray player should have two HDMI outputs. One for the video, and one for audio. The audio HDMI output should use the audio clock with blank video at some rate that makes sense using the audio clock. Using the audio clock would allow the jitter to be reduced.

Vendors could even design HDMI parts optimized for audio using the audio clock. If they did, the HDMI connection could have even lower jitter than SPDIF. This is because the audio clock could use the differential clock pair of HDMI. and since the audio is not encoded onto a single wire, this clock would get two benefits: Differential clock and dedicated clock wires.

This is so right on in every aspect that it ought to be engraved on every consumer digital product that attempts to work with any audio and video signals. The issues raised and their proper general solutions apply to PCM, DD, DTS "standard definition" audio design and - even more critically - the newer supposedly higher-quality "high definition" audio formats.

Like watching a horrific train crash in slow motion, it has been a depressing ten-year period as audio has gone from acceptable multi-bit PCM processing in laserdisc players, to DD/AC-3 laserdisc, to single-bit DAC's, to more-compressed and processed DVD software and hardware, to all-in-one processors, to increasingly careless and cheap signal data transmission, and now to HD formats that raise the basic flaws to new heights and claim to be "lossless" and thus better than ever.

Jim's issues and the HDMI solution, for example, are critical. Wish there were more of him around with a huge market base so the solutions could make their way into our components.

Yes, it would cost us all a little more money, but the results would be less like the compressed, frequency-distorted, discontinuous portable noise that we all think is acceptable and more like what real music and sound actually is.

Frustration >> off <<
:)

Are Beta units in users hands yet?

Hi Jeff: I believe so. It's a holiday weekend here in Uk though so the few who have them here probably wont be able to play with them until early next week.

Gordon

What is the process used to upgrade the firmware on the Radiance, and will it be contolled via the GUI to allow the user to pick and choose which features to add? Also will the user be able to roll back if the update has unforseen problems?

It is a simple painless application that runs to serial port on the scaler.

I don't think there is any performance benefit in not updating firmware completely. Perhaps at some point in the future when the thing is maxed out on fpga they will allow folk to choose what features to impliment but that's got to be a long way off and I don't even know if they'd do it that way.

Yes you have always been able to roll back Lumagen firmware. Lumagen updates also do not overwrite your system setup and configuration.

Gordon

In fact, audio is only in the Radiance as a competitive feature. In my opinion, video and audio should always be kept separate. I consider myself an audiophile, and it pains me to see that the only way to get uncompressed multichannel audio is through HDMI. With the extra jitter this incurs it is quite possible that this uncompressed stream sounds worse than current DD and DTS streams through SPDIF. As discussed earlier in this thread, HDMI will likely add jitter to the audio stream. An excellent amp can have circuitry to reduce this jitter to some degree, but this is usually reserved for very high-end equipment. It is better to have a low-jitter clock in the first place.

Since HDMI is the way these non-compressed formats are being supported, HD DVD and Blu-ray player should have two HDMI outputs. One for the video, and one for audio. The audio HDMI output should use the audio clock with blank video at some rate that makes sense using the audio clock. Using the audio clock would allow the jitter to be reduced.

This can be partially or completely solved by purchasing a simple (and inexpensive) HDMI splitter from MonoPrice. Route the video signals to the VP and the audio to a receiver or pre-pro. One can hope that upcoming audio solutions will allow for lipsync delay on HDMI audio sources. Given this cheap, simple solution, I personally think a lower-end Radiance model can forgo the audio processing.

It is a simple painless application that runs to serial port on the scaler.

Unfortunately it was never painless from me. Most PCs don't have serial ports now and I, like others, had problems getting my USB - serial adapter to work. I also have to unplug my Lumagen and transport it to my desktop (I don't have a laptop).

Upgrading via a USB drive would have been very welcome for me.

Martin

I have a USB to Serial adapter that I use to upgrade my Lumagen HDQ, works very well. No issues.

I got it to work eventually.

hmmm... just wondering, since the hardware is already finalized , is it possible for Lumagen to put some official high resolution pictures of the unit in their site ? ...

btw, thanks for including the image shift feature. really appreciated.

Unfortunately it was never painless from me. Most PCs don't have serial ports now and I, like others, had problems getting my USB - serial adapter to work. I also have to unplug my Lumagen and transport it to my desktop (I don't have a laptop).

Upgrading via a USB drive would have been very welcome for me.

Martin

Martin:

There is a very elegant and simple workaround...sfogg posted the URL before and unfortunately I am unbale to find it but basically he found a device that allows one yo upgrade their "serial connecting devices" (i.e. Lumagen Vision series) over a network in that one end of the device is a serial connection to the Lumagen/other device while the other end is a CAT-5 connection...very simple in that:

1. To upgrade all one needs is a network connection on their laptop -- seems like this would apply to everyone...

2. Or, even better, as I have my wireless router and equipment all stashed in the basement I would actually be able to upgrade my Lumagen from anywhere in the house...now that is user friendly!


SFOGG, CAN YOU PLEASE POST THAT URL AGAIN...MUCH THANKS.


HTH.

Joel,

http://www.avsforum.com/avs-vb/showthread.php?p=10075402&&#post10075402

Shawn

Martin,

"Unfortunately it was never painless from me. Most PCs don't have serial ports now and I, like others, had problems getting my USB - serial adapter to work."

What serial adapter do you have? IME Keyspans and Belkins work very well. If you have one of the no-name units you might try picking up a Keyspan if you are still having problems.

Shawn

Thanks guys I'll look into this. My serial adapter works now but it was far from plug and play at first. I have a LAN port in my HT room so this looks like a neat solution.

Martin

Joel,

http://www.avsforum.com/avs-vb/showthread.php?p=10075402&&#post10075402

Shawn

Shawn, thanks...very helpful.

My Onkyo TX-NR1000 made the same claim for a receiver. I fell
for all the propaganda about "future proofing" my home theater
as one would be able to get upgraded card modules and just
slide them in to replace existing one. Well after a couple of years
I finally got rid of the unit because Onkyo never released any
modules in the US-- just alot of promises as they focused
instead on releasing new receivers instead.

The HDP was upgraded as promised. In fact since I got my early release unit it has kept pace with three generations of DVDO products with out costing me anything extra.

I have the Integra version of the Onkyo you are talking about, its 50/50 if we will get an HDMI 1.3 upgrade for that. If Jim Peterson confirms the upgrade potential of the Radiance then you can pretty much consider it fact.

JP, so far we have a nice group interested in the possible blending feature.

http://www.avsforum.com/avs-vb/showthread.php?t=851150

28 will deffinatly buy if it has this and 10 would like to see it but would still buy. so that is a nice sales pitch. At say about $ 4000 for the radiance(this is a Guess) that is $ 112,000 in sales for sure !

So have you put more thought into it? Is it still Possible?

Athanasios

Joel,

http://www.avsforum.com/avs-vb/showthread.php?p=10075402&&#post10075402

Shawn
that is cool and they have a wireless one too! but at 200 bucks i'd pull it out of the rack to update.

Athanasios

While we are voting for features, I'd like to vote for an all black face plate or chassis. :)

This can be partially or completely solved by purchasing a simple (and inexpensive) HDMI splitter from MonoPrice. Route the video signals to the VP and the audio to a receiver or pre-pro. One can hope that upcoming audio solutions will allow for lipsync delay on HDMI audio sources. Given this cheap, simple solution, I personally think a lower-end Radiance model can forgo the audio processing.

This, unfortunately, doesn't solve the issue I was addressing. While it does provide two signals: one for audio and one for video, the audio will still have extra jitter.

The issue I was addressing comes from the fact that the audio needs to be encoded onto the HDMI wires using the video clock. They are essentially asynchronous to eac other. So the encoding of the audio on the video stream is imperfect at best, and will add jitter to the audio.

With a dedicated HDMI audio output from a player, the audio clock could be used in place of a video clock. You need to fake out the HDMI chips to believe its a video clock, so the clock would need to be multiplied into the HDMI acceptable range, but this can be done with little impact on jitter, verses trying to encode it onto the wires using an asynchronous video clock.

hmmm... just wondering, since the hardware is already finalized , is it possible for Lumagen to put some official high resolution pictures of the unit in their site ? ...

There have been some photos on this thread, but I will try to get some high resolution photos posted on lumagen.com early this week.

JP, so far we have a nice group interested in the possible blending feature.

http://www.avsforum.com/avs-vb/showthread.php?t=851150

28 will deffinatly buy if it has this and 10 would like to see it but would still buy. so that is a nice sales pitch. At say about $ 4000 for the radiance(this is a Guess) that is $ 112,000 in sales for sure !

So have you put more thought into it? Is it still Possible?

Athanasios

We have talked a bit at Lumagen about adding blending. What we believe is that if we do it, it would be a separately named product. This is for two reasons. We would need to have a different FPGA (EDIT: a different software-load not a different chip) for this product than the main-stream RadianceXD. Secondly-- even with all 28 buying ;) -- we would need to fund the development work for what we believe is not a huge volume.

While we haven't done the work yet, we think geting this to work exceptionally with the various gamma curves provided by different projectors may make the support level pretty high. We might be able to get by with as single control for the gamma in the blend region, but I think we might need a more parametric solution such as what we have for grayscale andf gamma now to get a precise result.

Still, we are not planning at this time to do a blending unit. Though, as you can see, it is under consideration.

While we are voting for features, I'd like to vote for an all black face plate or chassis. :)

There seems to be enough interest in this now that I will do a test run. Getting an all-black front plate is not the issue as I mentioned earlier. I will have to see how much work we need to go through to get the laser-etch or silk-screen done on the curved surface for the front panel text.

I will post when I have more info.

There seems to be enough interest in this now that I will do a test run. Getting an all-black front plate is not the issue as I mentioned earlier. I will have to see how much work we need to go through to get the laser-etch or silk-screen done on the curved surface for the front panel text.

I will post when I have more info.

Jim, sounds great...looking forward to a (hopefully) black RadianceXD...

Jim, sounds great...looking forward to a (hopefully) black RadianceXD...

You mean Radiance XDB :D

- Rich

This, unfortunately, doesn't solve the issue I was addressing. While it does provide two signals: one for audio and one for video, the audio will still have extra jitter.

I was thinking that, if the audio processor has a deep FIFO for lipsync delay and reclocks the data, jitter could be reduced as to be a relative non-issue.

http://www.avsforum.com/avs-vb/showthread.php?t=851150

30 people want tthe blend option so far, keep on voting guys!

Athanasios

I don't need blending but I need a CMS. Can we have a CMS poll?

Don't think we need a poll for CMS. I think Jim has already stated it will be a feature at a later date.

Shawn

Yup I think CMS should be expected.

Mark

Assuming memory serves me and CMS stands for Color Management System, then yes we will have this. This can mean different things to different people so some more details are:

Current Vision series color management includes:
- 11-point parametric grayscale and Gamma
- Black and contrast levels per memory
- Color and Hue (common in other video processors)
- Color and Hue offsets for Red and Green (not so common).

Radiance has:
- All the above
- 21 point Grayscale and Gamma (not sure if we will make this parametric).
- Primary Gamut control

The current Vision series can do an excellent job mitigating primaries that are way outside the SMPTE spec. The Radiance improves on this since it will be able to control the extent of Red, Green and Blue at the primary points, in addition to controlling the 75 percent color-bar points for Red, Green, and Blue independently that the Vision series already has.

Assuming memory serves me and CMS stands for Color Management System, then yes we will have this. This can mean different things to different people so some more details are:

Current Vision series color management includes:
- 11-point parametric grayscale and Gamma
- Black and contrast levels per memory
- Color and Hue (common in other video processors)
- Color and Hue offsets for Red and Green (not so common).

Radiance has:
- All the above
- 21 point Grayscale and Gamma (not sure if we will make this parametric).
- Primary Gamut control

The current Vision series can do an excellent job mitigating primaries that are way outside the SMPTE spec. The Radiance improves on this since it will be able to control the extent of Red, Green and Blue at the primary points, in addition to controlling the 75 percent color-bar points for Red, Green, and Blue independently that the Vision series already has.

You may want to have the 21 point be at fixed intervals with 3 - 4 parametric controls that can be set to be anywhere in the grayscale for fine touch-ups or smoothing out the rough spots. Ideally, these 3 - 4 controls could be targeted at individual primaries, and not just white overall.

JRP. I assume 11 pt means IRE windows of 0, 10, 20 yada yada, and 21 pt of 0, 5, 10, 15 yada yada. If 10 can be para, what woud stand in the way of 5 para? And does the para stay within the window? If the width is wider than the window, calibration coud be hell.

Anyway we can get a dynamic gamma control in there? eg, say the current gamma is 2.2, then there would be an adjustment, in perhaps 0.1 steps such that you could adjust the entire curve from the internal default, eg to 2.1, 2.0, 1.9, 2.5 etc. When I set up my ProHDP I had to hand calculate various curves, store them in memories using the 11-step gamma, and then flip between the memories until I found the gamma curve I liked. It would be incredibly useful if there was simply a gamma curve control which adjusted the entire curve in discrete steps.

Mark

Assuming memory serves me and CMS stands for Color Management System, then yes we will have this. This can mean different things to different people so some more details are:

Current Vision series color management includes:
- 11-point parametric grayscale and Gamma
- Black and contrast levels per memory
- Color and Hue (common in other video processors)
- Color and Hue offsets for Red and Green (not so common).

Radiance has:
- All the above
- 21 point Grayscale and Gamma (not sure if we will make this parametric).
- Primary Gamut control

The current Vision series can do an excellent job mitigating primaries that are way outside the SMPTE spec. The Radiance improves on this since it will be able to control the extent of Red, Green and Blue at the primary points, in addition to controlling the 75 percent color-bar points for Red, Green, and Blue independently that the Vision series already has.

so will this CMS "fix" the JVC RS-1 color issues?
Thanks

so will this CMS "fix" the JVC RS-1 color issues?
Thanks
It is unlikely to fix it completely given that it is an outboard box, but how well it will fix it has to do with too many contingent variables:
Can the JVC take a deeper bit depth signal than 8-bits?
How does it respond to that signal, esp. how much correction has to be applied?
How finicky is the user in question?

Just looking at the posted gamuts for the JVC, getting dead-on primaries may take more correction (lost bit-depth) than folks want. I would be optimistic that some form relief is feasible/usable from the Radiance, though.

Bill

jrp,

is the primary adjustment going to be scaled to CIE points (+-X and Y values for each of the three primary points?)

is the primary adjustment going to be scaled to CIE points (+-X and Y values for each of the three primary points?)
Not Jim, but I've played with this a bit. This would be easier to do in uv space, since the stepping is perceptually uniform. However, easier is not "easy". The geometry is straightforward to implement, but involved, and is only good at an ideal level (basically vector addition in uv space).

Since the reality can vary, you would need measurement gear to tell the Radiance what the coordinates really are (rather than plugging in some from a reviewer), and then you would still need to observe the actual results as you made adjustments. When the ideal model ultimately breaks down, you will then need to go to the "primitive" controls to make adjustments. In other words, you will never get around the need for a more traditional interface, and then whether the sophisticated view adds incremental value for the development cost is an open question.


Bill

Bill,

"Can the JVC take a deeper bit depth signal than 8-bits?"

According to tstites no. Someone was going to try and ask the designer of the RS1 that question though to double check that.

Shawn

JRP. I assume 11 pt means IRE windows of 0, 10, 20 yada yada, and 21 pt of 0, 5, 10, 15 yada yada. If 10 can be para, what woud stand in the way of 5 para? And does the para stay within the window? If the width is wider than the window, calibration coud be hell.

Both 11 and 21 point start at 0 and step in even steps to 100 IRE.

The parametric affects the point being adjusted, with the effect diminishing to zero by the next parametric points up and down (even if those points have been moved).

It is certainly possible to have the 21 point be parametric. We have not done the work and I was leaving the option open for us to do it either way. Still an open question, so if people want it to be parametirc, they should feel free to chiime in.

Anyway we can get a dynamic gamma control in there? eg, say the current gamma is 2.2, then there would be an adjustment, in perhaps 0.1 steps ...
Mark

Certainly this request has come up before. I will make sure it is on our "under consideration list".

so will this CMS "fix" the JVC RS-1 color issues?

It should fix the issues. As I mentioned the Vision series already does a good job of this.

The Radiance adds control for the x,y points for Red, Green and Blue and so should complete the job of correcting the color.

Will this be a full implementation of CMS such that the entire gamut is corrected or will it simply correct Red, Green and Blue at the measurement positions, ie 100% Blue, 100% Green and 100% Red? Also, will it be possible to correct secondaries - I don't even know if this is necessary once you've corrected the primaries, but let's say you fix green and blue but find that cyan is still missing it's secondary target position, would it be possible to fix cyan too and have it ensure that all the colours between blue and green were correctly adjusted as well?

CMS is one of the key areas that projectors are still not implementing that us high-enders are after, and it would be a large feather in Lumagen's cap to have a full implementation in the Radiance.

Also, CMS should be available on a per-memory basis such that SD and HD color spaces can be corrected for...

Cheers,

Mark

jrp,

is the primary adjustment going to be scaled to CIE points (+-X and Y values for each of the three primary points?)

We haven't done the controls for this, so I'm not sure exactly how we would set this up. It might be that the user adds color from the other primary, as such a control might be simpler.

I agree it would be nice to be able to measure the x,y points and enter the measured points and have us change x,y to compensate. I'm not sure if we could do this accurately enough for all displays.

What I have been thinking is to put a primary point up with a probe. Then the calibrator would add in a small amount of the other two primaries to move the point to the right location. Simpler to implement, should be robust to use, but it would make automation have to more be interactive than read x,y and then compensate (i.e. read probe, adjust the amount of the other primaries, read the probe, etc.).

Also, will it be possible to correct secondaries - I don't even know if this is necessary once you've corrected the primaries, but let's say you fix green and blue but find that cyan is still missing it's secondary target position, would it be possible to fix cyan too and have it ensure that all the colours between blue and green were correctly adjusted as well?If the primaries are correct, the gray scale is correct, and the color decoding is correct, there would be no additional need to correct the secondaries.

If I understand Jim's description of the Radiance's abilities, then it will have adjustments for all of these features.

BTW, I concur that a properly functioning CMS would make the Radiance truly unique in the marketplace and greatly enhance its appeal for a lot of potential buyers who would otherwise not be interested in just a switcher/transcoder/deinterlacer/scaler.

Thanks for the clarification, Tom - I wasn't sure but figured there was no harm in bringing it up.

Mark

Will this be a full implementation of CMS such that the entire gamut is corrected or will it simply correct Red, Green and Blue at the measurement positions, ie 100% Blue, 100% Green and 100% Red? Also, will it be possible to correct secondaries ...

Also, CMS should be available on a per-memory basis such that SD and HD color spaces can be corrected for...

Cheers,

Mark

The primary point would be where you make adjustment, but we would interpolate from there all the way down to black. So the entire gamut would be affected. The order for calibration would be:

- Set the display's controls to get as close as they allow.
- Use Radiance Gamut control to set primary point.
- Calibrate grayscale/gamma using the Radiance.
- Calibate color, Hue and their offsets using the Radiance.
- Check results and tweak if needed using the Radiance.

The gamut control, grayscale and gamma will be associated with an output configuration since they are intended to match an associated display. These are really independent of the input type since the primaries are at the extent of the range and "gray" is when Cr or Cb are both zero.

You will also be able to specifiy the input color space and we convert colors correctly to match the output color space. In other words a color from a SD source would be at the correct x,y location when viewed on an HD display.

The Color and Hue are per input sub-memory and can be used to correct for differences between sources independently for each input resolution.

We haven't done the controls for this, so I'm not sure exactly how we would set this up. It might be that the user adds color from the other primary, as such a control might be simpler.I wonder if this approach would be useful? I know that the Crystalio folks have struggled with implementing a true saturation control to fix just these sorts of problems and have been unable to implement a control that affects color saturation that does not also substantially affect color brightness.

So, for example, if you wanted to desaturate green by mixing with blue/red, would you not also substantially lower the brightness of green in addition to its saturation, since green is about 71% of reference white and blue/red (magenta) is 28% of reference white?

If I do a thought experiment and imagine a green field that has been completely desaturated without having its other properties affected, it would look like a field of 71% neutral gray. Wouldn't that be a better candidate for color mixing to achieve desaturation?

Jim:

Any reconsideration on having an HD-SDI input available, perhaps a "pro version" ?

I wonder if this approach would be useful? I know that the Crystalio folks have struggled with implementing a true saturation control to fix just these sorts of problems and have been unable to implement a control that affects color saturation that does not also substantially affect color brightness.

So, for example, if you wanted to desaturate green by mixing with blue/red, would you not also substantially lower the brightness of green in addition to its saturation, since green is about 71% of reference white and blue/red (magenta) is 28% of reference white?

If I do a thought experiment and imagine a green field that has been completely desaturated without having its other properties affected, it would look like a field of 71% neutral gray. Wouldn't that be a better candidate for color mixing to achieve desaturation?
The matrix math gets to be a little tedious, but it is technically possible. Unfortunately, this is a lot easier to do in floating point (e.g., in external software), rather than in integer math (e.g., what most consumer electronics use).

Also, for green, this is reasonably easy to do at a granular level since red and blue are together less than half the luminance of green. However, imagine the size of steps you have to take to correct blue while keeping luminance the same. Each unit of desaturating blue adds more than 10x the amount of light to be corrected via decreases in blue. In other words, you can probably correct green errors saturation errors relatively easily while keeping its overall luminance the same. Correcting red would be quite a bit harder, blue would be very difficult to do, unless you needed to make a huge change all at once.

Bill

The matrix math gets to be a little tedious, but it is technically possible. Unfortunately, this is a lot easier to do in floating point (e.g., in external software), rather than in integer math (e.g., what most consumer electronics use).

Also, for green, this is reasonably easy to do at a granular level since red and blue are together less than half the luminance of green. However, imagine the size of steps you have to take to correct blue while keeping luminance the same. Each unit of desaturating blue adds more than 10x the amount of light to be corrected via decreases in blue. In other words, you can probably correct green errors saturation errors relatively easily while keeping its overall luminance the same. Correcting red would be quite a bit harder, blue would be very difficult to do, unless you needed to make a huge change all at once.The math for this is completely beyond me, but based on my experience blue almost always requires the least adjustment and the human eye is least sensitive to errors in blue in any case. It's errors in red and green that are more critical to achieving a satisfying image.

... I'm just thinking, we are asking the unit to do something so complex (even just to think about it) , because , errh, one particular PJ model is doing the colors wrong .... ? I mean isn't this almost like the feature to do pixel misalignment "correction" ? Dun get me wrong, I'm all in for the color correction/calibration feature, but I think there is a limit this unit can do now given the limited time and resources. In the meantime please dun forget there are a lot of other essential features that are waiting to be implemented.

It is more than just one PJ "getting the colors wrong" - most PJ's have "incorrect" primaries. Most 3 chip PJ's have convergence problems. The issue for me is: what features would make it worthwhile for me to upgrade from the HDQ to the Radiance? I assume that the Radiance will end up having at least the features of the HDP line, so what additional things could be done? Primary color correction and convergence are big things that I would like to have. The whole point of a VP is to do these sorts of things.

Primary color correction and convergence are big things that I would like to have.It's also not just that these features are desirable. They are currently unavailable on any competing product. It would make the Radiance more than just another scaler among several very good ones already on the market.