Mark Haflich is going to be the host for a video processor shootout this weekend. He's bringing the DVDO VP50 to the table, Tom Huffman will be supplying a Vantage-HD, and I'll bring my unboxed Crystalio II VPS3300 which just arrived from Hong Kong today (and hopefully will not be DOA).

I'm going to leave it to the experts Mark and Tom to setup the criteria/tests for shootout. But it should be fun to have 3 different proprietary processors grinding away! :D

I'm going to leave it to the experts Mark and Tom to setup the criteria/tests for shootout. But it should be fun to have 3 different proprietary processors grinding away! :D

Dang - I'm hoping your sig is wrong and you live in Iowa :p - I'd love to see that. Willl be awaiting your results...

awesome! by any chance some Lumagen guys can join ? That will be the maga thread of the year!

The problem with adding a Lumagen is that the competitive deinterlacing that will be in the Radiance is not in the current models. We would be comparing old gen to new gens. When the Radiance comes out, that will be the time to do a shoot out that includes Lumagen. I can hardly wait to get one, hopefully as a Beta tester.

I really hope I get a replacement VP50 by Friday. Of course, if one doesn't mind the stripes, it is otherwise working fine. On a football green field, the stripes are barely noticeable. However on red and blue backgrounds, ugh.

My screen is a 4 way masked 1.78 110"D Stewart Studeotech 130. Room is totally light controlled amd black.

Please call me at 240 876 2536 if you are coming. I live at 22200 Zion Road, Brookeville, MD 20833. Brunch will be served. Come anytime Sunday morning and stay late. The lake we live on is really pretty this time of year. Lots of wildlife. There will of course be beer and soda besides the food. It won't be a party because we have serious work to do. But there will be good food and drink.

^^ What are the Lumagens lacking in deinterlacing compared to the other models mentioned? With recent firmware the Lumagens have per-pixel motion adaptive deinterlacing like the rest.

You guys please please please give us (the unlucky) some in-depth reports after your shoot-out. :)

Man, some people have ALL the luck... ;)

Mark,

I'm thinking about adding a VP at some point (soon?) and interested in observing this shootout. Since I live so dang close, I should make the effort. What time this Sunday are you meeting?

You posted your screen info, but didn't mention anything about a PJ or sources. I forget if you have a Ruby or something else. I know that you sell Sony, just not sure what you have in your HT. I have a Ruby, Toshiba HD-A1, and HD-DirecTV, so I'm most interested in seeing how much an improvement a VP would make in my HT. Let me know.

Here are a few thoughts about what you could do in the shootout:

(1) Deinterlacing tests, scaling should be turned off in the VPs here (should be done by display instead).
a. How good is SD film deinterlacing?
b. How good is HD film deinterlacing?
c. How good is SD video deinterlacing?
d. How good is HD video deinterlacing?

(2) Scaling tests, deinterlacing should be turned off in the VPs here (should be done by source device instead).
a. How good is SD upscaling?
b. How good is HD downscaling?

(3) Noise reduction and detail enhancement tests.
a. How good are the algs doing what they're supposed to do?
b. How big are the negative side effects?

(4) Deinterlacing + scaling + noise reduction + detail enhancement at the same time.
a. How good is SD film PQ?
b. How good is HD film PQ?
c. How good is SD video PQ?
d. How good is HD video PQ?

The problem with adding a Lumagen is that the competitive deinterlacing that will be in the Radiance is not in the current models. We would be comparing old gen to new gens. When the Radiance comes out, that will be the time to do a shoot out that includes Lumagen.
You can do the tests 1a, 1b, 2a and 2b with the Lumagen HDP. All other tests would be unfair. But the mentioned 4 tests should be a fair comparison test for the HDP, too. Especially a scaling only comparison (test 2) would be VERY interesting IMHO. I'd love to know how Crystalio II scaling compares to Lumagen scaling.

I wish I could come over for the shootout! Too bad it's a bit too far away from Germany... :(

P.S: Test (4) is especially interesting because AFAIK the Realta works differently compared to all other VP chips. If the Realta only does deinterlacing, it spends more processing power on the deinterlacing algs. If it's also doing scaling + noise reduction + detail enhancements, the Realta has less processing power left for deinterlacing. All other VP chips are running all algs totally independent of each other, AFAIK.

Madshi. Why not just turn it off and NOT have it done elsewhere?

Guys. I have a HIGHLY HIGHLY modified Electrohome 9500LC ULTRA. It does up to 1200p 72 hz, no sweat. It is a CRT. It has NO internal video processing (no internal scaling or deinterlacing) and it only accepts RGBHV in. I run it mostly at 1080p 59.94 and watch a lot of football (video). I do watch some films. My DVD is an old Denon (DVD 3300?) with SDI out.

I do not have a HD-DVD. I will be getting the Sony and Pioneer Elite BluRays when they come out. I can transcode any component to my long RGBHV run or any processor can do the transcoding.

The DVDO and Lumagens have no noise processing or detail enhancement etc. The new Lumagen wil have such features.

Madshi. Why not just turn it off and NOT have it done elsewhere?
Well, even better, of course!

However, you can't scale without deinterlacing first. So at least for a pure scaling comparison, all the VPs should get the same deinterlaced video stream and then apply their private scaling algorithm. Otherwise you can't really check the quality of the scaling algorithms seperately from the deinterlacing algorithm.

I would use a static test pattern to test the degree of ringing for the scaling. Sure, one would have to deinterlace from 480i to 480p first, but since there would be no motion, all the deinterlacers should be about the same.

I do not have a HD-DVD. I will be getting the Sony and Pioneer Elite BluRays when they come out. I can transcode any component to my long RGBHV run or any processor can do the transcoding.
Mark -- Do you want me to bring my HD-DVD player with an assortment of disks to the shootout as well?

Yes.

Mark- This is exactly what I need right now! A first hand critical assessment of what the best in video processing can do. I'm within striking distanced in PA, but need a pass from my wife to make it in person. I'll let you know closer to the weekend. My projector is a Panasonic AE-900 (720P LCD). Any need for one of these for the shootout?

Good job Mark, can't wait to hear the results..

BTW like you I am a HD sports fanatic so Video based processing of 1080I and 720P is of importance

Yes.

ok

I am sure this isn't going to go all that smoothly. Lots of different things to hook up, no set methodology etc etc. But we will have fun and learn how to do it better for the next shoot out.

If someone is coming from a distance, we do have a nice guest room for say Saturday night. We have had shootouts and meets at my house before. It's a nice place to visit and a nice place to live.

Hopefully I'm not stretching the limits by posting an answer to flyingvee.

For film sources, I would put the "old" Lumagen's against any of these boxes. We have proprietary "no-ring" scaling. No one else does (that I know of). We also support 24, 48, 60, and 72 Hertz output with correct cadence and stay in film lock exceptionally well. The Lumagen's also have a long list of calibration features.

The newer video processors do have some advantage for video (diagonal filtering) and noise reduction (temporal noise reduction for SD source for Silicon Optix based units verses our spacial noise reduction). These are not in the Vision series but will be in the Radiance series.

I personally look at film processing as my benchmark, but I know others prefer to look at video. You might want to consider us and our value compared to these other processors.

---

So Mark, while I know you don't like the SiI504's video deinterlacing, I would think that we rate being in your shoot-out.

Nice pj - would be worth a trip just to see your 9500LC. With your display there won't be qa whole lot of masking of limitations in the VPs - if it looks good on your rig, chances are it will be just fine on 98% of what is out there. And as you say, no internal scaling, dl, or any other processing - while methodology will probably be argued for weeks after, you should have an excellent idea of which processor does what by the time you are finished.

Tho I will have to agree with c722 - I'm curious about the current Lumagen's performance, if for no other reason than to see exactly how much the new units have progressed. I realize you would only want the very best for your rig, but for those of us with funky old 8" units, if the Lumagen were to give, say, 90% of what the new ones do, for a fraction of the price, that too would be of interest.

That is really great. Where else but on this forum. This will be appreciated by many.

Thank's Jeff

Darn - Sorry this is going to be Sunday. I have guests coming in from Fla. plus I need my VP50 for the Ravens game. Otherwise I would either attend or loan you my VP50 in case yours isn't working well. Instead, I'll have to watch the game on my C3X thru my VP50 and hope to hear MOST of the commentary. :D

For film sources, I would put the "old" Lumagen's against any of these boxes.

...and that includes 1080i60 "film" ?

C722: In answer to your question above....YES. As Jim says, the current Lumagen range can hold it's own with film source and can also do cadence locked 48,60,72Hz out for these sources.

Gordon

I've compared Lumagen, Crystalio and Vantage before and the Lumagen had 'em for cadence detection (especially 2:2 for us in PAL land) and is the better scaler IMO. Stick it in the test and be amazed at what a processor almost half the price of the others can do.

VP50 has a couple reported issues of bad cadence detection at least for 2:2 material over on AVForums. Since it uses any:any detection it may affect other material too. Might be worth holding back until a firmware update can be released for it (assuming the firmware update is going to help)

I already did my own very similiar shootout a couple weeks ago. It should be fun. With overall picture Q I already know where I would place my bet... :)

OK. Will include the Lumagen too. I have an older one but it's on loan to a customer whose dwin transcanner went south. JP doesn't know it yet, but he will be 2 day airing me one today.

Still don't have a replacement VP50 yet either. Couldn't get thru to their customer support yesterday. Hopefully its in transit.

VP50 has a couple reported issues of bad cadence detection at least for 2:2 material over on AVForums.To be fair, so does the Lumagen. In particular, users are reporting that it's a bit over-zealous in locking to a 2:2 cadence. That approach may work fine with film-sourced material, but with video the result is that it gets a lot of false 2:2 positives which cause combing and/or stuttering motion. Based on the user reports I've seen on the AVForums board, the VP50 is much better than the Lumagen processors in this respect.

By all means you should include the Lumagen in the shoot-out, though. It reportedly has excellent scaling and it should be interesting to see how it compares as it's substantially less expensive than the others. If I'd question anything, it would be the inclusion of the VP50 unit which is known to have a manufacturing hardware defect (assuming that Mark's replacement unit does not arrive).

- Dale Adams

I'm very pleased that we'll have a Lumagen on hand. I would have considered the event incomplete without it.

Any chance of adding in one of the newer less expensive VP to the comparison? I'm really curious to see how the new KeyDigital Isync holds up. HT Mag recently compared it to the VP30, but that was a while ago. Just a thought as I'm in the market for a new VP and of course would love to spend less.

I would try and get a Lumagen HDQ... Then you can compare the 1080i deinterlacing...

To be fair, so does the Lumagen. In particular, users are reporting that it's a bit over-zealous in locking to a 2:2 cadence. That approach may work fine with film-sourced material, but with video the result is that it gets a lot of false 2:2 positives which cause combing and/or stuttering motion. Based on the user reports I've seen on the AVForums board, the VP50 is much better than the Lumagen processors in this respect.

I agree with Dale on this one. I too have heard that we need a less-agressive film-lock mode for 50 Hertz video. With the current code we stay in film lock better than the other scalers (great for watching film), but we are planning a less agressive mode with a more video bias. There already is a "video only" mode for when the source is known to be video (=VIDEO for SD and either VIDNC or VIDPP for HD), but we are planning a new auto mode biased more toward video.

----

Mark:
We are sending a VisionHDQ out today by 2-day air, so you should have it in time for the shoot-out.

This is going to be great. Having owned the VisionHDP for over a year and the VP30 for a few months and now the VantageHD for 5 months, I think I know which I prefer :p I think Lumagen Scaling quality is still one of the best in the market. Its film deinterlacing is already right up there too. I know that if the HDP had video diagonal filtering, it would still be with me. I watch too much TV and sports to do without some sort of diagonal filtering. The calibration adjustments it has will also really help you dial in the perfect gamma and greyscale like no other processer can. Sigh.... I do miss my HDP

Now only if they hurry up with their new boxes. Looking forward to reading about this shootout.

Oliver

Nice one guys - should be a fun day. Remember us PAL guys ;)

Make sure it's the latest firmware on them all, as features and fixes are coming thick and fast to the new releases. A week could mean 48Hz on CII, Prep on VP50, etc, etc. (or it could not obviously).

Looking forward to reading all about it!

I get the replacement VP50 on Friday. Of course, I will have to switch out the SDI board from he defective VP50 to the new one. Only tkes a few minutes. SO we will have 4 video processors to test. The Crystalio, the Vantage, the VP50, and the Lumagen HDQ.

This is great as I have been thinking of picking up a processor. Will be waiting eagerly for your appraisal.

I get the replacement VP50 on Friday. Of course, I will have to switch out the SDI board from he defective VP50 to the new one. Only tkes a few minutes. SO we will have 4 video processors to test. The Crystalio, the Vantage, the VP50, and the Lumagen HDQ.

Nice list, the only thing missing in this list is a DragonFly.

I agree with Dale on this one. I too have heard that we need a less-agressive film-lock mode for 50 Hertz video. With the current code we stay in film lock better than the other scalers (great for watching film), but we are planning a less agressive mode with a more video bias. There already is a "video only" mode for when the source is known to be video (=VIDEO for SD and either VIDNC or VIDPP for HD), but we are planning a new auto mode biased more toward video.That doesn't sound like a bad idea. This has always been one of the problems with 2:2 detection - i.e., that it's possible to mistake true interlaced video for 2:2 due to the inaccuracies in correlating 2 fields which are not spatially coincident. And this has only been compounded recently due to the rising incidence of false 2:2 signatures induced by digital compression, where parts of the image which have little motion are updated every other field, while portions with a larger degree of motion are updated every field.

- Dale Adams

Nice list, the only thing missing in this list is a DragonFly.

The Vantage will have to stand in for the Realta on this one-- sorry. Its taken a long time just to get all of these processors into one place. All the dealers seem to handle one or the other and many don't carry the Crystalio or the Vantage.

Tom-- will you be loading the latest Vantage firmware before Sunday?

Lore

If any of you have suggestions/guidance on testing these devices methodically-- give us a post here.

Also-- If you have suggestions on DVD's or HD-DVD's you would like us to run-- please post the movie, DVD or HD-DVD, and specific chapter/scene that gives processors fits. I have about 40 HD-DVDs and a lot more DVD's -- so I may have what you are suggesting.

Unfortunately Silicon Optix new HD-DVD test disk is not ready and has not been released. We'll just have the DVD version.

Here are the links to the processor manuals. Take a glance through them if you there are particular features/settings you want to see tested and post.

Crystalio II VPS 3300
http://www.crystalio.com/downloads/Crystalio_Owners_guide_v1.1.pdf
--------------------------
DVDO VP-50
http://www.dvdo.com/documents/DVDO_iScanVP50_PG_ABT_081706.pdf
--------------------------
Calibre Vantage-HD
http://www.calibreuk.com/documents/Vantage_manual_121.pdf
--------------------------
Lumagen HDQ
http://www.lumagen.com/docs/VisionHDQ%20User%20Man.pdf

Doing this really well is going to be tedious and time-consuming.

I would suggest running a single test from the HQV Benchmark DVD and/or the test DVD that DVDO provides with the VP50 with one processor, change processor and then run the same test again until we've gone through all of the processors. Then, and only then, move on to the next test.

When we've finished looking at test patterns, then move on to actual program material using the same methodology.

- SD Video (DVD Concert Video)
- SD Film (DVD movie)
- HD Video (Broadcast HD video source, such as David Letterman and/or Jay Leno or HD DVD concert video, if one is available)
- HD Film (HD DVD film)

I'm looking forward to the results... thanks Guys! :D :D :D

I get the replacement VP50 on Friday. Of course, I will have to switch out the SDI board from he defective VP50 to the new one. Only tkes a few minutes. SO we will have 4 video processors to test. The Crystalio, the Vantage, the VP50, and the Lumagen HDQ.

Mark,

This sounds more and more like a really worthwhile adventure. If for some reason you don't receive your new VP50 as expected - or it is DOA, etc - let me know and I will still loan you mine. My guests aren't sophisticated enough :D to probably notice if I switch from my Oppo @480i thru VP50 to my Pioneer Elite 79avi for upconverting any DVD's and my cable HDTV will suffice direct to my PJ. I can do without it for a couple of days. I live in the Pasadena area and can pretty easily get my unit to you at the last minute - as long as it is before the Ravens game at 1PM on Sunday (prefer Sat/Sat night). Just let me know if you want or need it. I have no video/banding/line issues, only the audio dropouts. So it should be a good test bed.

Ray

Thanks Ray. DVDO dealer sales sent one to me today 2 day air, so I should have it no sweat.

Because Saturday is a busy sales day at my store, I am limited to doing events such as this at my house to Sundays.

Actually, I rather be watching the Redskins and Ravens instead of screwing around with processors but hey duty calls. I'LL be dVRring the games.

I'll try to push the Crystalio II to have the 3/2 pulldown fix mentioned in this post (http://www.avsforum.com/avs-vb/showthread.php?p=8131358&&#post8131358) before the shootout.

regards,

Li On

Looking forward to reading about the event (and the outcome!)

I'm really looking forward to this review too. If you guys want to fly out to San Diego and do it here I'll be happy to supply a couple of free donuts :)

Hi all,

It's kind of sad that I couldn't fly 20 hours to join the shootout party ... pls. keep the crowd posted of the results ASAP.

About the setup, I have a few suggestions:-

1. With a CRT, it should be fairly easy to test deinterlacing performance of all VPs alone. After this deinterlacing test, the winning VP(s) could then be daisy-chained via HDMI/DVI to the other VPs for up/down-scaling tests. This, IMHO, should be the fairest and should eliminate any ambiguity as to whether the artifacts are deinterlacing-oriented or scaling-orinted.
2. If at all possible, pls. include a test on the greatest looking combo - say VP1 for deinterlacing and VP2 for scaling - I'm thinking of keeping my HDP Pro as scaler and add a Realta-based or VXP-based VP as deinterlacer.
3. You guys may need a device to overcome the HDCP issue ... with the Toshiba HD-DVDP.
4. You guys will also need DVI-HDMI converters now that HDQ is in the test.

Just my 2 cents. Have fun.

:)

AK

We also support 24, 48, 60, and 72 Hertz output with correct cadence and stay in film lock exceptionally well.

Hi, thanks for posting here.
As far as I know, the HDP/HDQ is not able to output 24p yet, did I miss something ?
I know it's in progress on a future firmware update but not available yet.

Can you confirm ?

Thanks

SO we will have 4 video processors to test. The Crystalio, the Vantage, the VP50, and the Lumagen HDQ.
That is just perfect!

1x Gennum VXP
1x Realta HQV
1x ABT102 (HD)
1x Lumagen FPGA

:)

Guys. I suspect this will be just a shake down. Tom, Lore, and I live close to eacy other. 20 minutes or so with me in the middle. My house is always open. So when we start testing and realize we are missing some converter or cable etc, we will be better set up the next time. Also we have had no time learning the Crystalio. It takes a while to set each processor up correctly for different modes etc. This is going to be a learning session for all of us. Definitive results will probably await a later shoot out session.

That's a good point Mark. I had each processor for a couple of weeks before I new the ins and outs. It's not like you can just hook them up and make a good observation quickly... But I did know the CII was something very special the first hour after I had it hooked up and going...

Have you guys got a solution for HDCP? As obviously you'll be feeding an analog display?

Also don't forget that the Lumagen, VP50 and CII have 5xBNCs and the Vantage has D-SUB analog out.

I'll have to dig out my HQV benchmark disk:

http://www.hqv.com/benchmark.cfm

There is a written booklet that goes with it for evaluation.

For anyone coming to the shootout Sunday (including Tom & Mark), you may want to lookover this booklet in advance and print a copy out for your use:

http://www.hqv.com/contentEngine/dspDocumentDownload.cfm?PCVID=6557af58-7e90-e2a3-bea3-f6ec25bf8781

Have you guys got a solution for HDCP? As obviously you'll be feeding an analog display?

Also don't forget that the Lumagen, VP50 and CII have 5xBNCs and the Vantage has D-SUB analog out.

There wont be sufficient time to test all the outputs on all the processors. I guess we'll have a discussion before we begin what we want to try in this first effort. As Mark said, this may be a warmup--but I would certainly like to get some data for comparison.

We have to use analog out only. I do not think we need to bother with sound at this go around, There are lots of tests other than the Silicon Optix tests. Those tests are weighted to score the Realta high.

We have to use analog out only. I do not think we need to bother with sound at this go around, There are lots of tests other than the Silicon Optix tests. Those tests are weighted to score the Realta high.

OK on the analog out.

I know I have also heard those RUMORS about the HQV disk but it would be interesting to compare the processors on some of those tests anyway and see if the HQV really does come out first. I don't know of anyone who has ran the HQV disc through 4 processors like this side-by-side. Anyway it will be your and Tom's call. I'll just be the notetaker. :cool:

If you can only use analog out then with the CII you will only be able to do 1080i out. And since 1080i deinterlacing is one of the biggest concerns for many (if not then shortly it will be) then I would wait until you can do HDMI out comparisons...

If you can only use analog out then with the CII you will only be able to do 1080i out. And since 1080i deinterlacing is one of the biggest concerns for many (if not then shortly it will be) then I would wait until you can do HDMI out comparisons...
That's not true at all for RGBHV - 1152p is perfectly possible as I've done it.
All of these processors and even the last-gen ones could output 1080p over analog RGBHV.
The only limitation is HDCP in = no analog out at all. But I guess they've got that covered.

I understand what you are saying but for them majority of new owners HDMI inputs and outputs are the most important. I know the crystalio II's HDMI input 1 gets top priority. If HDMI is not a top priority for new VPs then why are they starting to come with 4 hdmi inputs and 1 or 2 outputs. HDMI needs to be looked at if you are going to do any kind of real comparison between these VPs.. Analog is going extinct...

Trust me - all of these can do 1080p60 output over RGBHV.
It's a non-issue.

DVDO - can NOT do 1080p over analogOh, come on. Even the old iScan HD could output 1080p60 on its analog output. The VP50 can output analog 1080p60 in RGBHV, RGBs, RGsB and YPbPr formats. In fact, the VP50 uses 300 MHz DACs and 2x oversamples its analog 1080p60 output to reduce potential problems caused by anti-imaging filters needed after the DAC. It uses 4 separate DACs, with one reserved for generating embedded sync, so that the dyanmic range of the active video signal is not compromised with RGsB and YPbPr. Do any of the others do this?

- Dale Adams

Really - for all my bitching, even I have had good results out of my VP30 and VP50 with 1080p output - have run at 48,60, AND 72. Works fine. Also have run at 1080p/60 over hdmi. No problems. If yours can't - something is broken, call DVDO for a new unit.

Guys - using the HQV Benchmark DVD test will be a start. Another great test tool would be the Stacey Spears and Don Munsil (now with a company name of Spears and Munsil) recently built test disk for evaluation of high definition processors. They are the creators of Anchor Bay Technologies new Video Reference Series disc. Might be worth seeing if the S&M disc is available to use as a test tool. At least many of the Test patterns on Stacey and Don's disc might be useful. However I am not sure if it is still in final beta or released.

guys great idea and I'm looking forward to your results.

About the HQV tests, these are cadence tests and I'm not really sure how they would refeclt in real life situations. They're great at comparing processors that have strong cadence squence detection, but comparing HDQ vs. VP50 vs. VXP vs. HQV on these will not really give you any usable results.

Real world tests with real world material (particularly 2:2 50Hz sources) are IMHO a much better test. The M&S disk is much more appropriate for such testing. If you can get a good 1080i source (TS content with a simple HDTV streamer will do and can output DVI without concern for HDCP & similar crap...)

I'm not really sure how you would compare the different items, I would suggest trying to prebuild (with our help) the test sequence as well as the rating system in each category. For example, something can have a strong scaling algorithm but a mediocre deinterlacer or vice versa.

Guys - using the HQV Benchmark DVD test will be a start. Another great test tool would be the Stacey Spears and Don Munsil (now with a company name of Spears and Munsil) recently built test disk for evaluation of high definition processors. They are the creators of Anchor Bay Technologies new Video Reference Series disc. Might be worth seeing if the S&M disc is available to use as a test tool. At least many of the Test patterns on Stacey and Don's disc might be useful. However I am not sure if it is still in final beta or released.
Pretty sure the VRS disc comes in the box with the VP50. Might have been with the ABT102 though, don't recall now.

Pretty sure the VRS disc comes in the box with the VP50. Might have been with the ABT102 though, don't recall now.

Both :D

I must have it somewhere at home then. We are simply not going to have this all together by Sunday. It will be a shakedown. We will do what we can and refine it and do it again later.

But my point is I would find the results over digital connections more important than analog. These days HDMI and DVI are top priority on most people's lists. I think it is great you are doing a shootout but without a digital assessment I would not be sure of the results. If someone is looking to buy a pj or tv these days HDMI is a lot more important than analog inputs...

But my point is I would find the results over digital connections more important than analog. These days HDMI and DVI are top priority on most people's lists. I think it is great you are doing a shootout but without a digital assessment I would not be sure of the results. If someone is looking to buy a pj or tv these days HDMI is a lot more important than analog inputs...

Truth. ;)

Great. But we won't be doing digital out. It is not going to happen this Sunday. I have a great CRT. It has no internal processing. It can take 1080p 72 and display it. Hell it can display 1200p @ 60. It doesn't care. I do have a 1080i digital server that I use in my store to feed Sonys. I'll bring it home. I have a Pioneer Elite BluRay (not fully functional) that outputs 1080p. I may bring that home too. But too much stuff for one day. This is not a one day project. This is my busy season. I am working 12 hour days at the store. I simply can't spend my life setting up this test. A little at a time. Later I can bring hoe a few digital; projectors. But they look so crappy compared to my CRT (remember my beast has been modified by the great Mike Parker, it is a large step beyond any CRT those here have ever seen).


Whatever. If I didn't have a very special CRT, I would have and be happy with several projectors out there. Three Sony's, the 004, the 100, and the 50. The new coming JVC DLA-RS1, A Digital Projections. The new Meridian. Lot's more. Good stuff out there but for now I have no need for any of them. Processors are a different story. I can't run my system with out one unless I want to use the crappy one built into my DTV DVR or built into some DVD player. The new Denon 5910 is real good, but I watch mostly sports on TV. Motion Adaptive Deinterlacing, world class scaling, and noise reduction and detail enhancement are what I want. Better and better processors keep happening and I will get them as they come out. Hell they all look great on film. But I am a video watcher. SD and HD.

Hey, just a reminder that the HDQ has DVI in/out - you need the right cables/adapters. Sorry if this has been taken into consideration already, I just want to hear the results (I have the HDQ), and I would hate to see the HDQ abandoned due to lack of connectivity during the shootout. Sounds like great fun! Wish I could be there.

I have HDMI/DVI and DVI/HSMI adaptors. Rarely sell them anymore, usually just order in cables with the right ends. The HDQ and the other stuff except the Vantage have SDI in. Plan on using that for DVD in. My 1080i server and the partially enabled Pioneer BluRay have both HDMI and component outs.

I can't wait to hear some results, I wish I could be cryogenically frozen until sunday!

That's the most absurd post I ever have read. I have heard about guys wishing they could get laid over a weekend with a special hot chick but never have I heard somebody wish they could be cryogenically frozen until three hobbyists reported results from testing four video processors. You be better off wishing for enough frozen marguerites to hold you until Sunday. Me ? I am praying for enough frozen marguerites to carry me through the testing Sunday. :)

Ofer. We gotta get to this guy. He needs help. I think the only thing that might cure him is a new video processor with motion compensation video processing. :)

The replacement VP50 and the Lumagen HDQ both arerived today. My pockets have two hdmi to dvi adaptors in them.

The replacement VP50 and the Lumagen HDQ both arerived today. My pockets have two hdmi to dvi adaptors in them.

Mark -- too bad you couldn't order PS3's and Wii's as easily....

Yea. Even Sony employees can't get the PS3.

Given the time constraint on this round, I'd recommend starting with actual movies in both film and video and if there's time, move on to test disk.

That way you also don't bias actual differences (or lack of differences) with normal content by test disk results.

Hooked up the VP50 and HDQ last night. Fed em both by SDI and set output to 1080p unlocked component. Right off the bat I liked the coded number call up system of the Lumagen rather than the toggle system front panel read out of the DVDO. But once you do the initial set up and can see the OSD on your display, they are both very workable.

JP. You know what would be nice. A large sheet print out of the commands for old guys like me who need to put on specs to read the manual.

Of course I had a test disc in the DVD player and the first tests were video and designed to flunk anything that doesn't have motion adaptive deinterlacing of 480i. As that famous statement went, a Sil504 deinterlacer under video (not film) tests is like a pair of brown shoes at a black tie function. Watching actual video stuff instead of tests is quite tolerable although because of my interest in VP I have become very sensitive to deinterlacing artifacts and immediately notice the artifacts that motion adaptive deinterlacing eliminate. Bring on that Radiance.

Miscellaneous musing. Now that projectors are including better and better deinterlacing chips, the main market for new VPs might become those with older projectors and CRT guys like me. Of course for me there must be a a high quality video DAC either in the chassis or available in a separate box. There are a several reasons why a separate box makes sense. First it keeps product costs down for those who do not need analog and it allows digital transmission until you get close to the analog projector input.

JP. You know what would be nice. A large sheet print out of the commands for old guys like me who need to put on specs to read the manual.



I 2nd this request. I have the HDP-Pro and would love to have a sheet with all of the menu commands in one place. They are scattered throughout the manual now, and I have my own cheet sheet of my commonly used commands, but would love a listing of all of them.
Is there an excel spreedsheet with all of them listed?

I just returned from the shootout and those who have been anticipating our results will likely be disappointed. I'm disappointed.

Because of the nature of the setup we were unable to do 2 things that I thought were essential for getting clear, reliable results: rapid switching between sources and testing of all four modes: SD video and film and HD video and film.

Our test PJ accepted only RGBHV and had only one input. Because of the complexity of swaping out processors it was simply unrealistic to look at one test pattern, switch processors and then look at the same pattern. Thus, we went through the entire HQV Benchmark disc with one processor and after swaping units we looked at the same patterns again. This meant that there was more than an hour between each test, and unless there were radical differences it was hard to keep track of relative performance.

We had setup problems getting the right colorspace out of units which was necessary before we could even get a menu to appear. We also had a problem with getting the Crystalio outputting a proper black level. By default, it was so low that we couldn't even see some of the test patterns. Because of all of these logistical problems it took us 5 hours to finish was I thought was an incomplete test.

Because of all of these time-consuming setup problems, we didn't get a chance to look at 1080i sources at all. This put the Lumagen at a severe disadvantage, because its strongest feature is with 1080i film sources.

Having said that, I do have some results to report, none of them terribly surprising. Lore and Mark can post their own results separately.

DVDO VP50
This unit performed very well on all of the test patterns with 2 exceptions. It failed the noise reduction tests and it took longer to lock onto the film detail test (the racetack) than any of the other processors.

Lumagen HDQ
The Lumagen really doesn't shine with these tests. It's weakest performance is with SD video sources, which this disc uses extensively. It performed well on the film cadence test and the film detail test. On the remainder of the tests it either failed or showed marginal performance.

Crystalio II
The Crystalio performed similarly to the VP50, with 2 exceptions. It faired well on the noise reduction tests and it locked onto the film detail test almost immediately. In fact, the only areas in which I felt the Crystalio had room for improvement were:
a) it needs to offer a better mix of noise reduction and detail. With the noise reduction turned on we saw very little noise, but detail seemed to suffer. When we turned noise reduction off, detail looked good, but it then had the same noise problems of the VP50 and Lumagen.
b) As I mentioned above, there was something strange going on with the Crystalio's black level performance. It seemed unusually low and unlike any of the other processors.

VantageHD
The Vantage performed similarly to the Crystalio with 2 exceptions:
a) It passed the jaggies test patterns but didn't do quite as well as the Crystalio (or the VP50 for that matter).
b) It did a better job than the Crystalio of providing noise reduction without compromising detail. In fact, noise reduction was probably the best area or performance for the Vantage.

I would like to repeat this test after the first of the year when the Radiance and HD test discs are available and focus on 1080i performance. Also, next time we must use a PJ that allows for multiple inputs, preferably HDMI, that we can switch between and do quick comparisons.

Also, next time we must use a PJ that allows for multiple inputs, preferably HDMI, that we can switch between and do quick comparisons.
Use a high quality analogue switcher so different projector input calibrations are irrelevant.

I would like to repeat this test after the first of the year when the Radiance and HD test discs are available and focus on 1080i performance. Also, next time we must use a PJ that allows for multiple inputs, preferably HDMI, that we can switch between and do quick comparisons.
Thanks for the report. :)

For HDMI switching, maybe the Zektor.

http://www.zektor.com/
Zektor - Home

Or, do any of these processors have unmolested HDMI passthrough? I think the the VP50 is supposed to but currently doesn't for 1080i sources. Might have HDCP issues trying that way though.

I'm going to watch football using the VP50 the rest of this evening.

I'll report tomorrow.

My only comments are that we had no experience with the Crystalio and spent some time setting it up and exploring its menu settings. There was obviously a problem with it that we could not solve. Probably operator inexperience with it.

Obviously if any of the processors lack noise reduction features (i.e., the DVDO and the Lumagen), they would fail any NR tests. What a surprise and a total waste of time testing them and reporting the results.

Obviously since the Lumagen does not do motion adaptive deinterlacing of 480i, again it was a waste of time testing nd reporting how it failed 480i motion adaptive deinterlacing tests.

Good night all. More tomorrow. I will get my hands on a component switcher so I can do quick comparisons. The differences in motion adaptive 480i deinterlacing between those units doing it (all except the Lumagen) were small.

Obviously since the Lumagen does not do motion adaptive deinterlacing of 480i, again it was a waste of time testing nd reporting how it failed 480i motion adaptive deinterlacing tests.Huh? The SiI504 used by the Lumagen does motion-adaptive deinterlacing of 480i and 576i. It may not do as good a job as you would like, but it does do it.

Now, if you wanted to claim that it doesn't have edge-adaptive processing (aka, diagonal interpolation or something similar), I could maybe buy that. (Although actually it does. If you look at the SiI504's processing on the HQV test discs's 3-bar diagonal test you can see that the top bar looks pretty smooth.) But to say that the Lumagen processor doesn't have motion-adaptive processing for 480i is simply incorrect.

- Dale Adams

Sorry Dale. I stand corrected. Obviously I am confused. I had thought the new generation chips, the Gennum, the ABT 102, the Realta, all brought motion adaptive deinterlacing of video to the table. Obviously, they all bring motion adaptive deinterlacing of 1080i film and video to the table. The Lumagen too does 1080i motion adaptive deinterlacing through in house developed algs and programming of a gated chip. Obviously, all the chips do a much better job of 480i video deinterlacing than the Sil 504. The Sil 504 seems to fail all the 480i deinterlacing tests on the HQV test disc. Is the improvement edge-adaptive deinterlacing, better algs, greater and quicker number crunching or what? Let's leave out cadence detection,detail enhancement, and noise reduction for the moment. Once again sorry to all for my misunderstanding. Live and learn or is it learn and live?

Sorry Dale. I stand corrected. Obviously I am confused. I had thought the new generation chips, the Gennum, the ABT 102, the Realta, all brought motion adaptive deinterlacing of video to the table. Obviously, they all bring motion adaptive deinterlacing of 1080i film and video to the table. The Lumagen too does 1080i motion adaptive deinterlacing through in house developed algs and programming of a gated chip. Obviously, all the chips do a much better job of 480i video deinterlacing than the Sil 504. The Sil 504 seems to fail all the 480i deinterlacing tests on the HQV test disc. Is the improvement edge-adaptive deinterlacing, better algs, greater and quicker number crunching or what? Let's leave out cadence detection,detail enhancement, and noise reduction for the moment. Once again sorry to all for my misunderstanding. Live and learn or is it learn and live?
I believe there are 2 main differences:

(1) The newer chips are better with cadence detection, bad edit handling and with detecting scenes where some parts are video content and others film, or when there are video overlays over film content. The older chips have more problems with this, as far as I understand it. This alone already explains why the newer chips do noticably better in some of the HQV tests.

(2) The edge diagonal processing is BIG TIME responsible for better video deinterlacing results. Especially for me, as I absolutely *HATE* jaggies.

I just like at add something about the CII tweaking for SD.

It is quite hard to manage this unit as I uses 2 chips. I also did the HQV tests on my unit and I found some big differences between the result given by the Fli2300 and the VXP. In some domain the Fli2300 is better, but in the other hand the VXP is alos better than the Fli2300 in others.
The CII uses a conditional managment called DynamicVP. This tool must be used in order to optimize this vp for SD materials.

Anyway, thanks guys for the shootout, which is a good job !
I just hope you'll get time to do it again and have less problemes.

Obviously, all the chips do a much better job of 480i video deinterlacing than the Sil 504. The Sil 504 seems to fail all the 480i deinterlacing tests on the HQV test disc. Is the improvement edge-adaptive deinterlacing, better algs, greater and quicker number crunching or what?It's been a long time since I ran the HQV tests on anything, especially a 504. It's not really one of my often-used test discs as many of the tests are for noise reduction, and the edge processing tests are a bit too artificial, short, or limiting. That disc is as much a marketing tool for SO as it is a true test disc, just as the ABT disc is for ABT.

Tests like the waving flag and rotating white bar(s) will not look good as they're designed to test edge-based processing, not motion-adaptive deinterlacing. The noise reduction tests are for, well, noise reduction. :) The Super Speedway bleacher scene, miscellaneous cadence, and mixed 3:2 tests are really tests of cadence detection, not motion-adaptive processing. I believe there's really only one test which which might be intended (partially) for motion-adaptive deinterlacing - it's the one which actually looks like a test pattern :D and has color bars, gray ramps, and some alternating black/white line patterns. And it's really only good for testing detection of no motion, as it's not a mixed motion/no-motion test pattern. (Some of the old Sage/Faroudja test disc clips are better for this.)

The Lumagen should do fine on this last test. What were your results for that pattern?

[Edit] The waving flag clip can also provide a test of basic motion-adaptive processing. The thing to look for is the detail in the bricks on the building - this should not be lost as there is very little motion in that area of the image. There should be no combing in the waving flag, but you'll need edge processing to reduce the jaggies caused by aliasing. One other thing to look for, and this is one that many observers overlook, is the stability of the stars in the flag. Older deinterlacers like Faroudja's DCDi can cause those stars to flicker or wobble a lot more than with some of the newer processors.

- Dale Adams

Well-- as the third person at the first shootout yesterday, I guess I should add my comments as well.

First -- thanks to Mark for hosting and the great snacks his wife made for us!

Second-- it was great to have Tom there as he's an ISF calibrator (an excellent one if you need someone in the VA/MD/DC area BTW)-- and as such lent an important methodical discipline and guide to our attempts to review the 4 processors.

Now -- my impressions. It is very, very, hard to evaluate these processors unless you can flip from one to the other almost instantly. As Tom said, the logistics in Mark's theater which were typical for any home setup didn't lend themselves to setting up all 4 processors and flipping between one to another with the touch of a switch.

Next-- the Crystalio had problems interfacing to the SDI input/analog output to Mark's setup. I reviewed the manual last night and there is no specific reason that it should have been as dark an image as it was. When I returned home I hooked the Crystalio into MY home theater setup (Tivo HD, HD-DVD, PElite DVD, HP MD6580 DLP, etc) and saw NO darkness problem when I inserted the HQV disk in the PElite DVD player set to output 480i over HDMI. ALL my home connections are HDMI right now. I will add two component in connections for my XBOX360 and Nintendo WII as soon as I get some RCA/BNC plug converters (Crystalio uses all BNC sockets) which I ordered from monoprice.com last night. So the problems with Mark's setup were NOT seen on my all digital setup. BTW-- the HP DLP takes 1080p60 for input.

I concur that even using the Crystalio on LOW for noise reduction still removes some of the sharpness. However, it does give it a more film-like look and the choice would certainly be in the eyes of the beholder.

I think it would be interesting to repeat the tests with the Lumagen Radiance in the setup. However I don't want to repeat on an analog setup (sorry Mark).

I want an all digital setup (perhaps the Sony Pearl or something better) next time. We also need a long table to setup ALL 4 processors at once, we need a 1x4 HDMI digital splitter to generate 4 equal HDMI streams into each processor and we need a 4x1 HDMI switch on the other side to quickly flip from one processor to the other. Having everything setup like in a lab is the only way to repeat this and not spend 90% of our time with the physical/wire movements. I have a switch but not a digital splitter that we could use. Unfortunately I don't have the space in my extra bedroom where my home theater is to accomodate the testing.

That's about it from here. I will say that all processors produced a nice image from the Superbit "The Fifth Element" DVD that was output over an SDI-interfaced Denon player. But that's not the test for dealing with "difficult" source material. And I was disappointed that we didn't have the time or energy to explore the HD-DVD player input.

I feel better about the fact I didn't manage to do this before I got C2 now :)
It's obviously a really really difficult thing to do without huge amounts of time, patience and equipment.

Really really appreciate the effort though guys.

I believe there's really only one test which which might be intended (partially) for motion-adaptive deinterlacing - it's the one which actually looks like a test pattern and has color bars, gray ramps, and some alternating black/white line patterns. And it's really only good for testing detection of no motion, as it's not a mixed motion/no-motion test pattern. (Some of the old Sage/Faroudja test disc clips are better for this.)

The Lumagen should do fine on this last test. What were your results for that pattern?

Dale:

As I recall, on this test the Lumagen did fine, as did all the others.

I'm new here, so sorry that I jump in with a question which may be strange, but as a more "scientific" approach where limited by the cirumstances, could each of you just give a subjective opinion on which VP you would prefer after this test?

Personally I trust the gut feelings of an experienced user more then all laboratory tests together.
Thanks
Stefano

I'm new here, so sorry that I jump in with a question which may be strange, but as a more "scientific" approach where limited by the cirumstances, could each of you just give a subjective opinion on which VP you would prefer after this test?

Personally I trust the gut feelings of an experienced user more then all laboratory tests together.
Thanks
Stefano

Hi-- I've owned the Vantage-HD for about 5 months, then the VP50 for a few days, and just go the Crystalio II. Its a little soon to pass judgement. I would say that the Vantage produced an outstanding picture on both 480i and 1080i sources-- but was fraught with constant firmware problems. Every time Calibre fixed something, something else "broke." I could never get all my sources working properly at the same time. My conclusion was that the Realta was a great processor that was virtually unprogrammable to the size or quality of the Calibre programming staff. And this may hold true for other companies producing Realta-based systems. They are just too hard for the end-vendor to program. I think Lumagen made an excellent decision to disgard the Realta for a VXP processsor in the upcoming Radiance product. No matter how great a processor is-- if you can't fix/expand/enhance its capabilities with a reasonable amount of man-months its not worth using it in your product.

I worry that the VP50 with is FPGA's may also be in the same boat. Mapping new algorithms (especially with Dale gone) onto their hardware may prove to be extremely complex-- leaving most of the upgrade potential and processing power of these devices essentially left untapped over their lifetime. And considering how complex it must be to debug-- heaven help them if their is a bug in the firmware.

For the couple of days I had the VP50-- I thought it produced an excellent picture from 480i HDMI output of my DVD player. I didn't really test it on 1080i material as it had a hardware flaw in some of its inputs and I wasn anxious to pack it up and exchange it for another. I found DVDO's customer service was OVERrated by alot of folks in this newsgroup. I waited over a week for an RMA so that I could ship it back and still hadn't gotten a replacement for over a month. I then decided to move onto the Crystalio II.

Since I've only hooked it up last night to my theater-- I'm going to need some days to evaluate it. I find the menu/user interface to be the best of all the processors but the proof is in the image. I did use it to deinterlace 1080i/60 recorded on my HD Tivo last night and the image was excellent. I also liked the image it produced from my Tosh HD-DVD player (Mission Impossible III). However the CII is reporting via its test screens that I'm not getting 1:1 pixel presentation on my HP MD6580 DLP set and this I need to explore before I can really test its quality. The question is - does it produce a picture worth the 30-40% price differential over the Calibre Vantage-HD??

I've post further subjective comments next week. Relatives are coming in for Thanksgiving and they want to watch tv not watch test patterns! :)

Hi-- I've owned the Vantage-HD for about 5 months, then the VP50 for a few days, and just go the Crystalio II. Its a little soon to pass judgement. I would say that the Vantage produced an outstanding picture on both 480i and 1080i sources-- but was fraught with constant firmware problems. Every time Calibre fixed something, something else "broke." I could never get all my sources working properly at the same time. My conclusion was that the Realta was a great processor that was virtually unprogrammable to the size or quality of the Calibre programming staff. And this may hold true for other companies producing Realta-based systems. They are just too hard for the end-vendor to program. I think Lumagen made an excellent decision to disgard the Realta for a VXP processsor in the upcoming Radiance product. No matter how great a processor is-- if you can't fix/expand/enhance its capabilities with a reasonable amount of man-months its not worth using it in your product.

[...]



Thank you for your report but calibre has delivered a very good firmware with 123A which is the last in date. No regression appears and picture is very good. In fact it is the best version until now.

I think that next firmware for vantage HD will be a great version.

Regards
alex_t

Hahaha, so your relatives don't want to see test-patterns? And I thought all the time that test-patterns where made also to free the home from undesired friends and relatives.:))

Thank you for you extended describtion. Debugging may be the real criteria how to select today an electronic device, but this info we will hardly get anyway. But to stay in the argument, the other day I went to read the PixelMagic forum, and it seems they also had a lot of problems with the last update 1.24 and apparently many users requested the older version back.

Currently I'm struggling between the Vantage and the Crystalio. Unfortunately in Italy when you don't like a product you cannot just send it back, so the decision making is harder.
So I wait for newer input after your relatives are gone, thanks
Stefano

Thank you for your report but calibre has delivered a very good firmware with 123A which is the last in date. No regression appears and picture is very good. In fact it is the best version until now.

I think that next firmware for vantage HD will be a great version.

Regards
alex_t


Alex-- I for one will APPLAUD Calibre if they have finally got their act together and fixed their software team's QA problems. As I said-- their unit produces and outstanding picture for both 480i and 1080i sources. If the operating system is free from gliches-- its a great product that I would recommend highly to anyone.

[BTW-- I still believe that that the Realta chip itself is a nightmare to program and no one has shown otherwise. The software development toolset that Gennum produces for the VXP processor is very user friendly. Does anyone want to show some screen shots of the development tool kit for the Realta???]

You guys are my heros. That has to be a pain in the butt, but you guys do it any way. There should be some Knighthood award for the work that you are doing.

My question is regarding a Plasma Display (i.e. Pioneer Elite Pro-1140HD). Will these processors have the same effect on a Plasma as it does on your projectors?

Our test PJ accepted only RGBHV and had only one input

A nice start guys - but dont give up on the testing using the Marquee. I agree that the HDMI testing is seen as more relevant to 95% of users, but as Mark already mentioned the CRT gives a transparency that a digital panel cannot achieve at all the different resolutions, making comparisons amongst the processors easier to see.

So , why not get yourself a high quality RGBHV switch so that you can swap the video processors over at the touch of a button? If all the output timings are set identically from the video processors then again it should result on the closest, most controlled environment for comparisons.

Oh yes and re the Marquee and C2 issue, did you have known good resolutions set up in the C2 for it? If you need some example settings then I can send them to you , just drop me a PM.

keep up the good work :)

I hope you do this again once the Radiance is released from Lumagen - as that processor seems more in the league with the others you've tested.

A nice start guys - but dont give up on the testing using the Marquee. I agree that the HDMI testing is seen as more relevant to 95% of users, but as Mark already mentioned the CRT gives a transparency that a digital panel cannot achieve at all the different resolutions, making comparisons amongst the processors easier to see.

So , why not get yourself a high quality RGBHV switch so that you can swap the video processors over at the touch of a button? If all the output timings are set identically from the video processors then again it should result on the closest, most controlled environment for comparisons.

Oh yes and re the Marquee and C2 issue, did you have known good resolutions set up in the C2 for it? If you need some example settings then I can send them to you , just drop me a PM.

keep up the good work :)

As an all HDMI user, its just not worth the time and effort for me to do analog testing. There is too much variability in having to reset the phase and sizing of the image when you shift from one processor to another. This would NOT provide an instantaneous shift but would require still a number of minutes of eyeball adjustments to the projector. With a 1920 x 1080p fixed pixel display accepting HDMI 1080p, we could shift instantly from one processor to another.

I will keep using the Marquee for my testing. I have access to a boat load of bulb projectors too but I insist on evaluating a stand alone processor without going into a projector or plasma that has a built in processor.

I warned everyone that the various test discs really don't mean much. It is so easy to score using one but the results for the most part are meaningless. If you a diagonnel filtering etc, you won't see jaggies except on the low angle of the rotor. What does it mean if one rotor gets to 14 degrees befor jaggies and the other 12. How can you tell exactly where. What does it mean. Certainly if a rotor has jaggies high it is not as good as one with much lower.
But if no diagonal filtering, there will be jaggies and one does not need a test to know it. Your picture will have em.

Different weird cadence tests. Not relevant unless you are going to use weird cadences. Anamine, games, not for me but certainly for others. Dig Vid cams, by all means run the test.

Bottom line. Play video and film, report the results. This takes time.

I will borrow a component switcher.

When an unbiased test disc comes out, I will use it.

The real test is watching video and film.

All did well on the film tests on the Silicon Optix test discs.

Personally, I want a processor with various noise reduction capabilities and detail enhancement. With NR, you will lose some detail and sharpness but the price is right when the screen is filled with noise. Any one characteristic, such as sharpness you could pick a winner. But PQ is a function of many things including getting ride of noise in the sky and in dark portions of the pictures and getting rid of mosquito noise when needed.



Last night I watched video 480i and 1080i using the VP 50. Very very few artifacts. Quite enjoyable.

Tonight I'll give the Lumagen a try but since the game is in 720p, I will mostly be testing its scaling to 1080p. I will try some 1080i on it later tonight. My tests like last night, all video. No film.

Lots of stuff to play with the Crystalio. Different types of NR and different settings. A fiddeler's dream come true compared to the VP 50.

Climbing this mountain can't be done in one day. It won't be easy either. Also HDMI must be tested too.

we all gotta pitch in and convince mark to get a good matrix switcher for future shootouts...

I'm really dissapointed, I was looking forward to:
1. 1080i deinterlacing comparison (which we haven't had so far from anyone)
2. Like Dale says and I agree, reliance on artificial test material is too problematic these days as it is tuned to give good results with very specific algorithms (or algorithm families). I now test mostly with a homemade DVD that I call the "Ofer torture test" and holds almost every PAL (no access to NTSC at all) sequence that has ever given me problems on any processor. Each time I add one test to the list as I route my TV viewing through my PVR...

Hi lorelevitt,

There is too much variability in having to reset the phase and sizing of the image when you shift from one processor to another. This would NOT provide an instantaneous shift but would require still a number of minutes of eyeball adjustments to the projector

I feel your pain :)

If the processors are all outputting *identical* signals however, then this should not be an issue, other than a brief resync for a second or so as the signal gets swapped over - no worse than HDMI. As Mark says, doing these tests with digital projectors means that you are testing not just the video processor itself, but the interaction between the standalone video processor, and the internal video processing in the digital display. This makes the test results less valid , as different digital displays of course have different internal video processing. Of course if the digital display being used happens to be the one that you already own (or are planing to buy!) then this is a good thing, but can you imagine the sheer number of source devices x viewing material x video processor x digital display x refresh rate/resolution combinations there would be :eek:

Mark, if you can get a high quality , high bandwidth RGBHV switch then you should be all set. Ofer, how about sending a copy of your test disc - it could become the official standard for video processing testing (much more useful than discs full of cadences that are hardly ever used!), and as a result a much more fair and meaningful test. If you want we can put it up on an ftp site..

David,

Wouldn't Ofer's test DVD be PAL?

Yes. He said it was pal. And I said earlier that I would get a component matrix switcher. I suspect I can get one for free or on loan from Key Digital when they learn its purpose.

I am sorry for the disappontment but I knew that the first session would be a shakedown.

Since the Crystalio does not output 1080p over component but does over RGBHV, it might be better to matrix RGBHV instead of component. After all, matrixing component meant using my MP5 to transcode component to RGBHV.

I was at my store until late last evening and didn't get a chance to turn on my HT.

Paying bills. Between Fri and Monday, about 30 came in.

Since the Crystalio does not output 1080p over component but does over RGBHV, it might be better to matrix RGBHV instead of component. After all, matrixing component meant using my MP5 to transcode component to RGBHV.
Yes, not to mention that standard 1080p timings are probably less than optimal for your PJ.
Do all of these VPs allow complete control over the RGBHV output resolution timings so that you can enter the same figures in each?

reliance on artificial test material is too problematic these days as it is tuned to give good results with very specific algorithms (or algorithm families).I saw no evidence of this.

The HQV disc revealed superior performance coming from the VXP processor and the DVDO processor relative to the HQV processor on some of the tests. In particular, both did a little better than the HQV unit on the rotating bar jaggies tests. The HQV processor did better on the noise reduction test. With regards to the remaining tests, they seemed to perform about the same.

If the claim is that the HQV tests somehow skew results in favor of the HQV processor, I'd like to see some real evidence of this. BTW, the test disc that DVDO provides with the VP50 contains many of the same tests!

The real problem with using a test disc is not bias but relevance. What I wanted to do, but which we did not have time to complete, was real-world comparisons with actual SD and HD video and film program material. I'd like to see whether a processor's ability to perform better on a benchmark test is clearly reflected in real world improvements in picture quality.

The rotatimg bar is I think an unbised way to test for diag filtering and to what low angle the filtering alg and number crunching will work. The relevance of a few degrees difference in angle is probably none re world world viewing. However, the difference in algs will show up on how smoot the lines become. But once again relatively small differences. if a processor does do diag whatever, it will look pretty pooron this test as well as it will have jaggies on many video edges.

the cadences on the HQV disc are designed to match the cadences in the Realta. Of course the test is useful for determining if the other processors handle those cadences.

I'm sorry but the disc is set up for tests that the Realta will pass.
this does not mean other processors won't do as well on some of the tests.

The cadences are standard cadences used by variety imaging technologies unrelated to the Realta chip. Again, I'd prefer some evidence over mere assertion.

Actually I have done extensive 1080i deinterlacing comparisons between the VPs... And of course using the HDMI ins and outs...

And? You big tease!

I believe joerod is referring to the thread he started.

http://www.avsforum.com/avs-vb/showthread.php?t=746512

He is a big fan of the Crystalio.

My opinions though are 100% unbiased. I use what I like best and quite frankly the crystalio II is the best on the market... It just has the total package... :)

I saw no evidence of this.

If the claim is that the HQV tests somehow skew results in favor of the HQV processor, I'd like to see some real evidence of this.

Are you really that naive? The evidence is pretty apparent to me. HQV fails a lot of tests on the VRS DVD but passes almost all of them on the HQV DVD. The VP50 passes almost all of the test on the VRS DVD but fails some on the HQV DVD. Why? Hmmm...what a mystery. Lets see could it be because both companies want their processing to look its best and their respective marketing departments made sure that was the case when these DVDs were made?

The real problem with using a test disc is not bias but relevance. What I wanted to do, but which we did not have time to complete, was real-world comparisons with actual SD and HD video and film program material. I'd like to see whether a processor's ability to perform better on a benchmark test is clearly reflected in real world improvements in picture quality.

Agreed. Now you're talking. Yes, that would be the ultimate since nobody sits around watching test patterns....well, then again this crowd just might.

The cadences are standard cadences used by variety imaging technologies unrelated to the Realta chip. Again, I'd prefer some evidence over mere assertion.

OK, watch the hockey test scene on the VRS DVD with your HQV processor. It falls short of the VP50 and DCDi by a mile. Interesting that the HQV DVD doesn't have a hockey test scene on it though, isn't it? That is because the HQV marketing types knew their low angle processing wasn't as good as DCDi at the time they made that DVD.

The VP50 passes almost all of the test on the VRS DVD but fails some on the HQV DVD.This claim may or may not be true, but it is completely irrelevant to my post. The VP50 passed ALL of the tests on the HQV disc, except noise reduction. In fact, it performed somewhat better on the rotating bar jaggies test than the Realta processor. That it failed noise reduction is not terribly surprising, since the VP50 contains no noise reduction technology.

I asked for evidence "that the HQV tests somehow skew results in favor of the HQV processor." Your responded with name-calling and irrelevant comments about the VRS DVD, a disc about which I offered no opinion. Very helpful.

We looked at the DVDO-supplied disc briefly, but rather quickly determined that the HQV disc was a better tool for the test. Since the results from using it did NOT show the Realta processor was the clear winner, but rather revealed a variety of strengths and weaknesses of the 4 units, I am confident we made the right choice.

Exactly. Each test by itself has some validity and maybe some relevance, then again maybe not.

Sure as fecal matter, each company's test disc contains a collection of tests chosen to make its chip look good. Its the sum that's biased, not each test.

Ah relevancy. If a processor doesn't have NR, all noise reduction tests are irrelevant.

I am not going to participate in this argument. Dale is an expert and he says both discs are biased. Ofer is an expert, he says the HQV disc is biased. I am just an idiot, prone to mistakes in an area I love but do not completely understand. But logic is logic.

I like Tom and Lore, both good guys. But I disagree with Tom on this.

I have a great microscope, my FP CRT which contains no processing. Testing because of timing issues will not be easy. But hell during a HDTV broadcast with commercials, the phase shifts. I adjust when I change channels and go own. It ain't that big of deal.

I have conclusions but we have a mixture of apples and oranges. I want a processor with the best video algs for normal video cadences. I want NR and detail enhancement. Chips keep getting better. I can hardly wait to get a processor with the new Gennum chip, the Radiance and maybe the next Crystalio. I like all the tweaks on the present Crystalio. But its algs are already somewhat obsolete. The VP50 needs NR and detail enhancement to be competitive with what will come soon. They will come because the boys at ABT are shrewd. They want a big piece of the chip market and won't get it with MNR, general NR, detail enhancement etc.

And guess what, many projectors will be incorporating the latest chips. The real market for new processors will be front and rear sets without the latest, and me and the other CRT boys.

The Realta chip has proven to be difficult to work with by those who have used it. However, don't expect them to so state because their NR and license agreements prevent them from so saying. But I bet those companies with processors incorporating that chip rue the day they decided to do so. Not because the chip is bad, it is just a bitch to bring to market. And duh that market is small and limited. Looks like a money loser to me.

A processor manufacturer must monitor the problems its customers are having and make fixes almost on a continual basis. Lumagen has done a great job at doing this and continually adding features its customers want. It is not in the chip selling business. It is in the processor business. So is Pixel Magic. It chose to wait to bring out its product. It does a lot and doesn't have a lot of bugs. However it is expensive and the chips it uses are almost obsolete. The ability to plug in a new job as the chip manufacturers keep making their stuff better. Lumagen's strategy is very smart here. I expect Nelson at Pixel Magic, a great guy by the way, will follow suit.

The people using the Realta, Algolith and Calibre and a few others, have a tough row to hoe. It is not easy to keep adding things and very substantial investments have been required to get them where they are now. Improvements have been and I suspect will be slow. Calibre has pretty much announced they are about done. I bet they are sorry they ever undertook the project. it ain't gonna be a money maker.

I know most of the major manufacturers rather well. And I like all of them. I am routing for all of them because they bring products to market that I really enjoy and make my HT better. I will do anything I can reasonably to help them. But when I see what I think is marketing bullshit, I will point it out. My testing will continue. All present products are good. They could and for the most part will be better.

So is Pixel Magic. It chose to wait to bring out its product. It does a lot and doesn't have a lot of bugs. However it is expensive and the chips it uses are almost obsolete

Not sure I would agree with that Mark, sure it is at the top end in terms of price, but it has the most features. Regarding almost obsolete? well as Lumagen have shown the value is in the implementation of the chips at your disposal, these are software driven products.

The ability to plug in a new job as the chip manufacturers keep making their stuff better. Lumagen's strategy is very smart here.

..or very optimistic. Look at the PC marketplace as an example., how many different generations of CPU sockets are there? If you design something to be generic, then you compromise, increase production costs and development complexity, if you maximise the hardware and software then you get the best performance out of that package.

It just doesn't seem realistic to try and build a design that will predict and have the bandwidth and internal interfaces to deal with future chip designs that don't even exist yet, again just look at the PC mainboard marketplace as an example of this. These sort of statements always sound great and give potential users a 'warm feeling' , but of course nobody knows whether this upgradable chip theory works until we see the chip designs in 2007/2008.

time will tell...

If you design something to be generic, then you compromise, increase production costs and development complexity, if you maximise the hardware and software then you get the best performance out of that package.

It just doesn't seem realistic to try and build a design that will predict and have the bandwidth and internal interfaces to deal with future chip designs that don't even exist yet, again just look at the PC mainboard marketplace as an example of this. These sort of statements always sound great and give potential users a 'warm feeling' , but of course nobody knows whether this upgradable chip theory works until we see the chip designs in 2007/2008.

time will tell...

David-- I fully agree. I remember the early period of the Pentium processor when alot of folks wanted to upgrade their old processor to the Pentium. Many companies sprung up with add-on cards containing a new processor but the corresponding changes in the bus, hard drives, graphics cards, etc. did not make sense to only upgrade a portion of the system.

Similarly in the CE industry where technical progress has been somewhat rapid in the past decade. As protocols such as HDMI 1.3 and other changes occur, simply being able to change the core processor over a card-based bus will not produce the same high-end quality that the type of person buying a video processor is looking for. On paper it seems very appealing-- offer an "infinite" upgrade path to the latest and greatest from GENNUM. In reality these things have never proven out to really benefit the consumer.

One really good example was the Onkyo TX-1000 Receiver. This received great reviews in the media, cost $5K+ and with a bus architecture and slide out cards was supposed to be the end-all in receivers. Never buy a new one again-- just buy the replacement cards. Well what happened in reality is that Onkyo never released any upgrade cards. They would promise HDMI 1.2 than later say they were going to wait until 1.3 was standardized. They promised XM radio or Sirius on the tuner card. They NEVER considered True HD or other non-lossy audio methodologies. And years passed and they still DID NOT RELEASE A SINGLE CARD IN THE US!! I wish the folks building the RADIANCE would study the failed engineering model of Onkyo on this receiver. And keep in mind-- Onkyo has HUNDREDS of more engineers than Lumagen.

What good is the next generation of video CPU when we have seen that the complexity of the current one is more that the VP manufacturers can already handle. No one outside of Silicon Optix appears capabile of easily programming the REALTA architecture. I predict that the same complexity problems will befall the FGPA's used by DVDO and that true algorithm progress on that architecture will be slow or negligible. I don't know much about GENNUM's architecture on its chips but I will say I'm impressed on the surface by the high quality development tools they tout (VIPER application builder) for working with it:

http://www.gennum.com/ip/pdffiles/vxp_brochure.pdf

I've seen no such development environment described for the REALTA.

These chips are far more difficult to program for than programming in assembly code. Parallelizing and optimizing one's code remains quite difficult.

well as Lumagen have shown the value is in the implementation of the chips at your disposal, these are software driven products.
That depends. The FPGA chips used by the VP50 can be reprogrammed by a new firmware. The Realta runs software. However, the Gennum VXP chips are more hardwired than VP50 and Realta, AFAIK. There's only so much you can do to change how it works. E.g. Gennum has not been able to add full Anime cadence detection into the chip used by the Crystalio II. And this chip will never see full Anime cadence detection. I know that for a fact. So when using the Gennum VXP, it's not unimportant for a standalone VP manufacturer to be able to change to a newer chip without too much hassle.

The new VXP chip which will be used in the Radiance adds SD and HD compression noise reduction as well as improved random noise reduction algorithms. None of these algorithms will ever work on the older chip used in the Crystalio II.

In short: The Crystalio II in its current form will never be able to do compression noise reduction nor Anime cadence detection and there's nothing PixelMagic can do about that. They will have to redesign their motherboard to make use of a newer VXP chip, as the newer chip is (to my best knowledge) not pin compatible to the older one.

..or very optimistic. Look at the PC marketplace as an example., how many different generations of CPU sockets are there?
That's why Lumagen has chosen not to use a video chip socket, but a small daughter card. This way Lumagen is able to define their own interface. I guess they have designed it so that there's one video stream going to the video chip, one coming back from it and then there's probably one control logic. If they designed it cleverly, upgrading should actually work for a while. Of course we'll have to wait and see. But looking at the past, Lumagen has fulfilled most of their promises (although some only with big delays).

If they designed it cleverly, upgrading should actually work for a while. Of course we'll have to wait and see. But looking at the past, Lumagen has fulfilled most of their promises (although some only with big delays).

I don't care how clever the few designers are at Lumagen. Not even Gennum knows what their architecture will be in 5 years. Perhaps they will go in the direction of Intel/AMD and produce dual core or quad core video processors. Lumagen will NOT be able to produce an upgradable architecture beyond ONE video cpu generation. Now is all the added complexity worth it for just one upgrade-- I don't think so when you consider what the cost of that daughter card will be. I would prefer a generous trade-in arragement for a new box instead when a new cpu comes out.

Perhaps we should archive and time capsule this thread and go back and look at who was right in 2011?? :D

I don't care how clever the few designers are at Lumagen. Not even Gennum knows what their architecture will be in 5 years. Perhaps they will go in the direction of Intel/AMD and produce dual core or quad core video processors. Lumagen will NOT be able to produce an upgradable architecture beyond ONE video cpu generation. Now is all the added complexity worth it for just one upgrade-- I don't think so when you consider what the cost of that daughter card will be. I would prefer a generous trade-in arragement for a new box instead when a new cpu comes out.

Perhaps we should archive and time capsule this thread and go back and look at who was right in 2011?? :D
But look: How long does it usually take until a new VP with a new chip revision is released? Usually much longer than a year! Hell, even first announcement of a new VP and first actual shipment sometimes takes longer than a year these days! When do you expect Crystalio III to arrive?

For me the daughtercard stuff is not mainly about money. It's about getting new chips implemented as soon as possible. Normally, if Gennum releases a new chip, we'd have to wait months or years for a VP making use of the new chip. With the Lumagen daughtercard idea we should arrive at a very low month number. Provided everything works as expected, of course.

You're right that Gennum probably doesn't know for sure how their chips will look like in 5 years. But you should be aware that they're mainly targetting at projectors and flat panel displays. So I don't think dual chip architecture will ever be used. And I do believe that Gennum does have detailed plans what to do in the next 1-2 years. And I expect that Lumagen has some (limited) access to these plans. Do you think Lumagen has pulled the bit about "Video processing can be upgraded to 16-bit when 16-bit depth video processing chip becomes available" out of thin air? I never had expected a 16-bit depth video processing chip in the near future. I guess that Gennum told Lumagen that such a chip is on the roadmap. And I'm quite certain that Lumagen got enough information so that their daughtercard will be able to make use of it.

Normally, if Gennum releases a new chip, we'd have to wait months or years for a VP making use of the new chip. With the Lumagen daughtercard idea we should arrive at a very low month number. Provided everything works as expected, of course.

but that is just a theory.:)

think about it, you yourself said that rather than just implementing a socketed design, they are building a daughterboard. So, that daughterboard, by definition, has more 'stuff' on it to support that chip, and it is now separated from the main board. That mainboard now has less 'stuff' on it, and between the two, a high speed, 'future proof, data bus has to be designed, built and tested. The design of this bus is critical as it has to work not just with the mainboard and the current target processor, but it also has to work with new unknown processors as well. So, lets just assume now that this costs extra R&D time for the first Radiance release over a fixed processor design. When the next VXP chip comes along, then presumably the other VP manufacturers will have complete reference designs to work from, whereas Lumagen will have the additional workload of building a new daugherboard that has to work with an *existing* released, fixed hardware design mainboard and bus. Will that be quicker than the other VP companies who will base their products on the chip makers reference design? nobody knows. Will it be worth it for the small production runs that video processors have (relative to other consumer electronics), who knows. We can take an educated guess though by looking back at history, and I don't know of consumer electronics product (or indeed any product!) where the core 'engine' can be swapped between product generations. Electronics is all about component reduction and rationalisation to reduce costs, increase performance and reliability, and that just doesnt fit with this approach.

I don't want to sound 'down' on Lumagen, it is a brave and laudable approach, but it seems a huge gamble to me, as at some point you have to tell your customer that their video processor investment that they purchased on the basis that it was 'future proof' is no longer going to be upgradable. Furthermore, even if it is upgradable, will the next generation chip work better when it is run through this bus and 'legacy' main board, or will it work better in a completely new, dedicated mainboard design? I think we can all guess the answer to that one!

HOWEVER, getting back on topic.

Mark, you said

Testing because of timing issues will not be easy. But hell during a HDTV broadcast with commercials, the phase shifts. I adjust when I change channels and go own. It ain't that big of deal.

You really need to get this sorted :) . The whole point of a video processor is that it should be insulating you from these changes, you should not be seeing the signal timings change during adverts!. To move forward you are going to need to go RGBHV to solve this. As we have all said, get yourself running RGBHV with a decent switcher, and you should have no problems with this, not to mention better results from your CRT overall.

See this is what happens when you take on a test like this, you get all sorts of random demands as to how you should set it up :)

all the best

Not sure I would agree with that Mark, sure it is at the top end in terms of price, but it has the most features. Regarding almost obsolete? well as Lumagen have shown the value is in the implementation of the chips at your disposal, these are software driven products.



..or very optimistic. Look at the PC marketplace as an example., how many different generations of CPU sockets are there? If you design something to be generic, then you compromise, increase production costs and development complexity, if you maximise the hardware and software then you get the best performance out of that package.

It just doesn't seem realistic to try and build a design that will predict and have the bandwidth and internal interfaces to deal with future chip designs that don't even exist yet, again just look at the PC mainboard marketplace as an example of this. These sort of statements always sound great and give potential users a 'warm feeling' , but of course nobody knows whether this upgradable chip theory works until we see the chip designs in 2007/2008.

time will tell...

I agree with this. Complexity is added without any guaranty that it will be used. I prefer the DVDO approaches where each upgrade bring you a new clean platform without legacy bottleneck.

That depends. The FPGA chips used by the VP50 can be reprogrammed by a new firmware. The Realta runs software. However, the Gennum VXP chips are more hardwired than VP50 and Realta, AFAIK. There's only so much you can do to change how it works. E.g. Gennum has not been able to add full Anime cadence detection into the chip used by the Crystalio II. And this chip will never see full Anime cadence detection. I know that for a fact. So when using the Gennum VXP, it's not unimportant for a standalone VP manufacturer to be able to change to a newer chip without too much hassle.

The new VXP chip which will be used in the Radiance adds SD and HD compression noise reduction as well as improved random noise reduction algorithms. None of these algorithms will ever work on the older chip used in the Crystalio II.

In short: The Crystalio II in its current form will never be able to do compression noise reduction nor Anime cadence detection and there's nothing PixelMagic can do about that. They will have to redesign their motherboard to make use of a newer VXP chip, as the newer chip is (to my best knowledge) not pin compatible to the older one.


That's why Lumagen has chosen not to use a video chip socket, but a small daughter card. This way Lumagen is able to define their own interface. I guess they have designed it so that there's one video stream going to the video chip, one coming back from it and then there's probably one control logic. If they designed it cleverly, upgrading should actually work for a while. Of course we'll have to wait and see. But looking at the past, Lumagen has fulfilled most of their promises (although some only with big delays).

Yes but to build the motherboard and daughter board socket, they are still making assumption on what the architecture will be in2-3 years and what the bandwidth will be and these may be wrong.

When the next VXP chip comes along, then presumably the other VP manufacturers will have complete reference designs to work from
There was a reference design for the Realta. But I've not heard of a Gennum reference design yet. To my best knowledge there is no such thing, as Gennum is not targetting standalone VPs, but mainly projectors and flat panel displays.

I don't want to sound 'down' on Lumagen, it is a brave and laudable approach, but it seems a huge gamble to me, as at some point you have to tell your customer that their video processor investment that they purchased on the basis that it was 'future proof' is no longer going to be upgradable.
Well, even if there were only 1 or 2 updates available, that's better than none.

Look: Buying a VP today is a bit risky. You get no HDMI 1.3 and so no support for DeepColor. That means the VP has to go down to 8bit RGB before passing the data to the display, which may result in banding, dithering or posterization problems. Actually I've read opinions of ISF calibrators which recommend not to use external VPs because of this very problem. Tomorrow's displays will have DeepColor and may even make use of the added color depth. If you buy a Crystalio II, you'll forever be limited to the current Gennum chip and you will never get HDMI 1.3 - unless you update to the Crystalio III, which might not even be in the planning stages right now (I've no idea). If you buy the Radiance, you'll be able to upgrade to HDMI 1.3 and you'll be able to update at least to a Gennum chip which will be able to maintain a full DeepColor path through the whole processing chain. Basically by choosing the Radiance you get around buying "too early" and you can get the latest technology retrofitted. Ok, I've no idea whether upgrades of the processing chip will still be possible in 2008. Maybe not. But at least I know that I'll be able to upgrade the Radiance so that it takes full advantage of DeepColor. That's something, isn't it?

Furthermore, even if it is upgradable, will the next generation chip work better when it is run through this bus and 'legacy' main board, or will it work better in a completely new, dedicated mainboard design? I think we can all guess the answer to that one!
I don't think the dedicated mainboard must necessarily work better. It depends a lot on how the daughtercard interface is designed. I don't have enough information about which communication channels the Gennum chip needs to say anything concrete about this.

Well guys--to paraphrase what I said earler..'those who fail to remember history are condemned to repeat it.' I think the Onkyo TX-NR1000 modular design is a excellent case of the failure of indefinite upgradeability.

We can go back and forth about arguing is 1 generation upgrade worth the extra engineering. How would the cost vary if a trade-in was done instead of an upgrade? There really isn't any answer to all this-- its a personal issue on a case by case basis.

I have no idea what and when a CIII will exist. I think that even when new chips come out, Lumagen will have a long time bringing out these daughter cards--much longer than you or they believe. Hell, Calibre STILL has not released either their HD/SDI option card (over 8 months since the unit was released and over a year since it was shopped around as shipping last fall) NOR their HDMI option card to which they have virtually refused to comment publicly on when that will be released-- if at all now. And I have to tell you-- these cards are ALOT simpler than the daughter card that a new video CPU would sit on.

I am wary of proprietary “upgrade slots” in anything, most of them have turned out to be failures in the sense that they never released anything to go in the slot, or what was released later was of no value due to the rapid evolution of better platforms. There is a way that makes sense to me to do “upgrades”, that is high-speed serial interfaces as seen in PCI-e slots on computers. The upside is a very high bandwidth connection, the downside is that you need more glue logic on the upgrade card. It has been stated in the past that the core scaling and processing chips of scalers are a very small part of the overall cost of the total device, and I love the idea of a fairly inexpensive way to change out the core and keep the I/O,PS, and other misc hardware. I just don’t know if it is worth it to the manufacturers to do this – it costs significant engineering time to abstract system cores. If people were not obsessed with HDMI 1.3,1.4b,1.666 it would be easier to keep a common chassis based on HDMI 1.1.

Can I just butt in here and point out that you are all making assumptions about how Lumagen are implimenting their multi board design in RADIANCE. I'm all for speculation but you seem to be damning something that you have very limited information about. Why not wait till one surfaces and have a look inside and see what the seperate boards all do. Then you may have your answers ;)

Gordon

Can I just butt in here and point out that you are all making assumptions about how Lumagen are implimenting their multi board design in RADIANCE. I'm all for speculation but you seem to be damning something that you have very limited information about. Why not wait till one surfaces and have a look inside and see what the seperate boards all do. Then you may have your answers ;)

Gordon
:rolleyes:

Gordon it is a lot more fun speculating on how it is done than to look at actual pictures.

Let's apply your reasoning to sex. It's a lot more fun speculating on how its done than looking at pictures. Based on that, your next logical step would be to say its a lot more fun looking at pictures than doing it.

Remember, Lumagen historically has used its outside chip to do deinterlacing of 480i and 576i. Almost everything else it does with its own algs programmed onto field prgrammable gate arrays (FPGAs).

JP has stated the Radiance will be packed full of FPGAs. While he might initially use some of the non deinterlacing features of the new Gennum chip such as NR etc, he has left lots of room in those gate arrays for procesingalgs Lumagen may develop in the future. JP's attitude has always been that Lumagen can do it better. Now it is very captal intensive to dvlop deinterlcing algs. Much cheaper to rely on someone eses and focus your energies on providing what your end users want. To allow for future things not even thought of yet.

Soos. Since others are dvloping deinterlacing improvemets, it makes sense to TRY and development architecture that will allow upgrading deinterlacing as chip manufactures such as Gennum, Silicon Optix, and ABT develop it. A bonus is ifother features are contained in the new chips it also can be used. Best of both world's.

Why no Realta chip initially. No inside infobut let me guess. Very high cost and a real bitch to implement. But dooable.

Let's apply your reasoning to sex. It's a lot more fun speculating on how its done than looking at pictures. Based on that, your next logical step would be to say its a lot more fun looking at pictures than doing it.
Actually the situation is...
You can't do it.
So - should we be talking about it, or are we just going to get all hot and bothered with no payoff :)

Mark,
Sorry I didn't know about the shootout. I completely missed this thread on the forum. Thanks Tom for telling me about it. Count me in for round 2; since I'm in Rockville it is not far for me to drive. I just got a VP50 and was, in fact, busy on Sunday (19th) setting it up. I have a Key Digital analog switch (the one I got from you, Mark) which I could bring to next shootout if you don't get one. I also have an Altinex, but it is only RGBH (4 analog connxs) and I don't think that would work. For HDMI I have several Gefen switchers, but am awaiting latest firmware upgrades. Their previous ones have problems. I have a Toshiba HD-AX1 and hope to get the newest model as soon as they release it to reviewers. I have a ton of HD DVD discs and thoughts on a few that would be good tests. I also have some HD DVD tests from Stacey Spears that I burned to DVD. I could also bring boxes of cables, adapters, etc. just to be sure we had everything needed on hand.
-Bill

The Realta chip has proven to be difficult to work with by those who have used it. However, don't expect them to so state because their NR and license agreements prevent them from so saying. But I bet those companies with processors incorporating that chip rue the day they decided to do so. Not because the chip is bad, it is just a bitch to bring to market. And duh that market is small and limited. Looks like a money loser to me.

A processor manufacturer must monitor the problems its customers are having and make fixes almost on a continual basis. Lumagen has done a great job at doing this and continually adding features its customers want. It is not in the chip selling business. It is in the processor business. So is Pixel Magic. It chose to wait to bring out its product. It does a lot and doesn't have a lot of bugs. However it is expensive and the chips it uses are almost obsolete. The ability to plug in a new job as the chip manufacturers keep making their stuff better. Lumagen's strategy is very smart here. I expect Nelson at Pixel Magic, a great guy by the way, will follow suit.

The people using the Realta, Algolith and Calibre and a few others, have a tough row to hoe. It is not easy to keep adding things and very substantial investments have been required to get them where they are now. Improvements have been and I suspect will be slow. Calibre has pretty much announced they are about done. I bet they are sorry they ever undertook the project. it ain't gonna be a money maker.

I know most of the major manufacturers rather well. And I like all of them. I am routing for all of them because they bring products to market that I really enjoy and make my HT better. I will do anything I can reasonably to help them. But when I see what I think is marketing bullshit, I will point it out. My testing will continue. All present products are good. They could and for the most part will be better.

Does anyone think it's possible that because of the difficulties with the REALTA chip that Calibre and Algolith may cede the high-end of the market to Gennum/ABT and make video processors based on the REON VX chip instead?

I think there is a large untapped market of people who are very disappointed about the quality of SD and even HD on their new HDTVs. The problem is that most of the video processors are built for the projector market and are very expensive.

Most 40" - 57" plasma/lcd owners have no need for all the various inputs, cadences, etc. They just want to plug something in that makes SD watchable and eliminates all the digital artifacts.

I think that Algolith especially should create a "plug and play" processor based on the REON VX chip. They have already taken the same approach with the Flea which costs 1/3 the price of the Mosquito. They even market the Flea as "plug and play". It is my opinion that if they could create an identical product that de-interlaces and scales everything to 1080p, then they will have a winner on their hands.

Thoughts?

As I understand it, and I could be mistaken, Silicon Optix has an ownership interest in Algolith.

Today I watched the Dallas game in NR 720p into both the DVDO VP50 and the Lumagen HDQ and scaled by those units to 1080p 59.94. Two observations.

First both units presented a great picture but in comparison the scaling by the Lumagen was superior. Cleaner and sharper. Another nice touch by Lumagen and needed by me for the comparison was an included set of RCA to BNC compoent cables BNCs are a much better choice for video connections than RCAs. Component cables are 75 ohms and so are BNCs. RCAs are 50 ohms and the mismatch can cause ringing. My CRT microscope can discern these differences.

Second on various commercials deinterlacing artifacts were often noticeable. Obviously neither unit was performing any deinterlacing. So I conclude the artifacts were in the commercials, either initially produced in 480i or 1081i and deinterlaced rather poorly and scaled before broadcast.

I will repeat later in watching video in NR 1080i letting both units deinterlace to 1080p. Stay tuned.

Second on various commercials deinterlacing artifacts were often noticeable. Obviously neither unit was performing any deinterlacing. So iIconclude the artifacts were in the commercials, either initially produced in 480i or 1081i and deinterlaced rather poorly in production, and then scaled to 720p.I've also seen quite a few SD deinterlacing artifacts in both 720p and 1080i signals. Since the artifacts were at SD resolution, they were obviously in the source itself and were not introduced by the video processor's deinterlacing stage. I've observed these on both commercials as well as 'real' program material.

- Dale Adams

I had the HDQ for a little while as well. I have to admit it held its own against the VP50. It is hard to justify the bigger price difference between them. I still think the HDQ was as good as or even a little better than the VP50... Of course the VP50 is a little more conveinant with its HDMI ins and out. But judging pic Q alone the HDQ can hold its own...

Last night I watched one of the poorest quality SD sources I have even seen, the NFL network broadcasting the Denver/KC game. What was awful was the broadcast quality of the game, the commercvials etc were much more viewable.

I initially used the Lumagen but the picture was so awful (fuzzy, unsharp, noisy etc) artifacts were buried in noticeability by the awful sourece. I switched to the DVDO and things didn't improve. While there may have been less deinterlacing artifacts (hard to tell because the picture again was awful), sharpness was a little worse (really the picture was so poor with either procesor it was hard to really judge anything).

I think some NR and picture sharpening would have helped but I was too tired to put my Algolith Mosquito into the chain.

That is a good point. That is another reason why the crystalio II got my vote. The NR features come in handy in situations like that... Even if they are very far and between it is good to know you have something that can help "tackle" the problem...

Spent some more time tonight watching college foot ball in NR 720p scaled by the Lumagen HDQ and VP50 to 1080p. The Lumagen was sharper, crisper on edges. Helmuts for example were much better defined along the edges.

Then I noticed mosquito noise along the edges of lines on the Lumagen. Mosquito noise was not visible on the VP50. The reason I believe was the noise was masked by the less sharp edges on the VP50. I surmise that the edges were ringing slightly on the VP50 and this blurring so to speak was masking the noise.

With MNR I think the Lumagen would be the clear winner. The Radiance will have MNR because it is available in the Gennum chip and I believe it will be added to the VP50 or the next generation DVDO. At that point, DVDO will need to improve or at least retune its scaling. WTF do I know, just my thoughts. Right now I would choose the Lumagen's scaling even with untreated MN.

Spent some more time tonight watching college foot ball in NR 720p scaled by the Lumagen HDQ and VP50 to 1080p. The Lumagen was sharper, crisper on edges. Helmuts for example were much better defined along the edges.

Ten I noticed mosquito noise along the edges of lines on the Lumagen. Mosquito noise was not visible on the VP50. The reason I believe was the noise was masked by the less sharp edges on the VP50. I surmise that the edges were ringing slightly on the VP50 and this blurring so to speak was masking the noise.

With MNR I think the Lumagen would be the clear winner. The Radiance will have MNR because it is available in the Gennum chip and I believe it will be added to the VP50 or the next generation DVDO. At that point, DVDO will need to improve or at least retune its scaling. WTF do I know, just my thoughts. Right now I would choose the Lumagen's scaling even with untreated MN.

Same result as me when I tested both the HDP and VP30/ABT102 upscaling 480i/P to 720P. I agree with your test result totally as well.

Oliver

I have a VP testing rig here. It consists of several SDI players (6 typically wired up at any one time). These feed a Kramer mechanical switcher and then a SDI distribution amp (1 in 4 out). This can feed 4 VPs simultaneously. I have these connected with DVI (Zektor switchers) for digital and RGBHV (Extron switcher) for analogue. This allows pretty quick testing when fed from the same source. It is just the time for the CRT to lock onto the signal. RGBHV is seemless and relatively quick. HDMI is more complicated as often there are HDCP ‘sync’ issues to deal with together with the issue that some VP have either analogue or digital outputs as opposed to both on at the same time. The DVI switcher feeds a DVI to RGB interface box which also goes into the RGBHV switcher and seems to deal with HDCP. This is all assessed using a CRT projector. This is basically a dumb ass device with no internal processing so gives a good idea about what each VP is up to. Seems to work well for my interests and is relatively quick to change but involves quite allot of cabling and volume of boxes. It also has DVI and component switchers that can be brought into action when comparing various different sources.

@Nic, it's nice to hear about your setup. But you forgot to post your opinion about how all your VPs compare... :)

Btw, have you finally succeeded in adding a SDI Philips 963 to your SDI player collection?

Madshi

I only occasionally post opinions on shootouts of VPs as they are highly dependent on what the end 'customer' wants. For me I have little or no interest in video and I am much more interested in HD and 48 capabilities of VPs than say noise reduction etc. I therefore generally limit my comments to specific requests for help though I obviously find it fun playing at home. The rig and my playing is largely for my interest.

Re 963, if you look at AVForums I think you will see I was the first to have / use a 963 SDI and was one of the early 'adopters' of this player. I actaully have three here currently, two SDI'ed and one even has an audio mod as well ;)

Re 963, if you look at AVForums I think you will see I was the first to have / use a 963 SDI and was one of the early 'adopters' of this player. I actaully have three here currently, two SDI'ed and one even has an audio mod as well ;)
Hmmm, sorry, I think I have confused you with somebody else. There's somebody on these forums collecting various SDI players, who didn't have a 963 yet.

Hi Nic
Have you had chance to see a HD-SDI source yet? SKY HD or HD-DVD player?

Today I watched the Dallas game in NR 720p into both the DVDO VP50 and the Lumagen HDQ and scaled by those units to 1080p 59.94. Two observations.

First both units presented a great picture but in comparison the scaling by the Lumagen was superior. Cleaner and sharper.

.

I can't take you serious as you seem clearly biased towards Lumagen to whatever costs.

I have done the same comparison but with a VisionHDP and the Lumagen is a far worse product on SD compared to the VP50 and with HD the VP50 cleary beats the Lumagen.

As i said earlier Lumagen fails in every department due to it's aging hardware.

Using words like "superior" does not make you credible at all, it's far from superior.

It's only superior in your own mind.

Hi Likvid - I think you misread the post.

The test done by Mark is purely to test SCALING - nothing else. He's starting out with a 720p signal so your comments about SD might well be true in your instance, but they certainly can't be used as an argument against these findings.

This is one person stating their experience and posting their method. I don't see how you can accuse him of bias at all from this.

I can't take you serious as you seem clearly biased towards Lumagen to whatever costs.

I have done the same comparison but with a VisionHDP and the Lumagen is a far worse product on SD compared to the VP50 and with HD the VP50 cleary beats the Lumagen.

As i said earlier Lumagen fails in every department due to it's aging hardware.

Using words like "superior" does not make you credible at all, it's far from superior.

It's only superior in your own mind.
Look, there are several things a video processor does. The 2 most important things are deinterlacing and scaling. And there are two totally different areas of deinterlacing, namely video and film deinterlacing. Your post does not differ at all between all these totally different parts of a VP. It is well known that the VP30 with ABT102 and the VP50 are noticably better at video deinterlacing. However, many people (Mark is just one of them) think that the Lumagen holds its own ground on film deinterlacing and many people (Mark is just one of them) claim that the Lumagen has a better scaling algorithm. So if you're not agreeing with Mark, that's just fine, but you should say in which area you're not agreeing. If you say the Lumagen fails in every department then most people here will not take you seriously because you're alone with that opinion.

Hi Nic
Have you had chance to see a HD-SDI source yet? SKY HD or HD-DVD player?

Not with SkyHD or HD DVD but I did have a small play with HD SDI a year or so ago, just to get ready ;)

Likvid. I hope your observing skills are better than your reading skills. :) Just kidding. My test was with HD. Native Rate 720p from Direct TV. I scaled it to 1080p with both processors and observed. My microscope was a 54 x 96 inch screen illuminated by a highly modified 9500LC ultra.

With my set up, the Lumagen's scaling was clearly superior. You could see Mosquito noise from the Mpeq compression and decoding. This was masked by the less clear and less sharp scaling done by the DVDO. I think the DVDO scaling has a moderate degree of ringing and this ringing along the edges somewhat masks things. I am not saying that the scaling by the DVDO is poor. No, it is good but the Lumagen's is better and noticeably so in my UNBIASED but not so humble opinion.

With my set up, the Lumagen's scaling was clearly superior.Since we are on the subject of scaling and these are the very two VP's I am considering, do you think the Lumagen would also be best for scaling the 480p output from a Nintendo Wii to 1080p? Right now the kid's GameCube connected via component to my Sony 70XBR2 shows severe jaggies on all diagonal lines (I'm talking about even static images without motion).

Would the Lumagen smooth this out adequately or would I need the latest diagonal interpolation algorithms of the VP50 or the Realta based VP's? Or is the superiority of this diagonal line processing refer to just interlaced video?

I suppose I should first ask, is the output from a Wii more like film or like video?

Jaggies come from deinterlacing, not scaling. So I think what you are seing is poor deinterlacing coming out of the Wii. In this case, the VP50 Prep feture might serve you well, in essence it would reinterlace the 480p signal to 480i and then the processor would deinterlace it without jaggies to 480p and then scale it to 1080p.

Jaggies come from deinterlacing, not scaling.
I haven't seen the Wii, but jaggies at 480i out of a PS2 is inherent in the source (no antialiasing), not a deinterlacing artifact.

He said the jaggies were visible when outputting 480p from the Wii. I would agree if the jaggies were visible when outputting 489i.

Jaggies come from deinterlacing, not scaling. So I think what you are seing is poor deinterlacing coming out of the Wii. In this case, the VP50 Prep feture might serve you well, in essence it would reinterlace the 480p signal to 480i and then the processor would deinterlace it without jaggies to 480p and then scale it to 1080p.While that's true for the most part, you can get rough or jagged edges from the scaling process, particularly when scaling a low resolution image up to something much higher resolution. In this case, what you're typically seeing are resolution limitations of the original source. I often see this sort of thing for instance, when scaling 480p up to 1080p. It's also true that some scaling algorithms make the original source limitations more visible than others.

To be a bit picky, jaggies generally come from aliasing caused by interlacing (although they can also be present in a purely progressive source, particularly those generated by computer). If the image is sampled such that a single field does not meet sampling rate requirements (i.e., the vertical sampling rate falls below Nyquist rate requirements), then you get aliases in that field. The aliases most often are visible as jagged edges on objects, as those edges are where the vertical high frequencies are most prominent. Edge-based deinterlacing algorithms, such as the well-known DCDi, are really intended to hide aliasing artifacts along edges, where the human visual system is very sensitive to them.

- Dale Adams

Blue 911, PReP would offer nothing in your case, the full frame 480p coming out of the Gamecube or Wii should not need to be re-interlaced as it was never interlaced in the first place. The better scaling of the vp50 or Lumagen compared to the screen will probably help and of course, you can use it for other sources too.

Rough edges I can understand from simply having low resolution to start with. Jaggies I think are different, being in my mind back and forth lines conecting points in a progressive image where there has been motion. Not a great explanation, but I think you know what I am trying to say. Roughness, rather than back and forth lines that I think we all call jaggies, is like a curve made by trying to conect too few points to make a smooth curve.

Now in such a case a scaling alg that say caused some ringing could blur the edges hiding detail but also hiding such things as the roughness and mosquito noise. I essence, this to me is a loss of resolution. I'd rather have sharp and crisp edges and solve the other problems by some other solution.

Dale. I appreciate your explanation. You obviously know more about this than almost anyone including mself who might know less than most.

Below is a screen shot from the Game Cube. The box says this game is "progressive compatible" and I am using the digital AV out port so I assumed it was 480p, but when I actually checked the TV input, it says 480i. So, Mark, you're right this may be a deinterlacing problem.

The static part of this image--the Zelda title--is scaled nicely, no jaggies on the diagonal lines. The dragon boat, which is moving around, has jaggies. Can someone explain this?

Not sure why the Sony is not deinterlacing properly. Would either the Lumagen or VP50 fix this?

http://i93.photobucket.com/albums/l61/ChuPhoto/GameCubesample.jpg

Rough edges I can understand from simply having low resolution to start with. Jaggies I think are different, being in my mind back and forth lines conecting points in a progressive image where there has been motion. Not a great explanation, but I think you knowwhat I am trying to say. Roughness, rather than back and forth lines we I think call aggies, is like a curve made by trying to conect too few points to make a smooth curve.
What you describe as jaggies sounds to me more like combing. This occurs when data is combined from 2 fields which have motion occurring between them. (It is possible to get a jaggies type of effect from this when the rate of motion is just right, but combing is the more common artifact.) That's definitely a deinterlacing artifact, while jaggies can simply be an artifact inherent to interlaced scanning. Jaggies, as I understand the term anyway, is a well-defined staircase type of effect, where an object edge is clearly stepped or jagged looking. It's not a back-and-forth type of thing, but more like this:

*************
*************
**********************
**********************
*******************************
*******************************

Deinterlacing without edge-adaptive processing might blur those sharp steps a little, but they would still be clearly there.

Combing along a diagonal edge would look like this:

**********
*******
***************
************
********************
*****************

- Dale Adams

The static part of this image--the Zelda title--is scaled nicely, no jaggies on the diagonal lines. The dragon boat, which is moving around, has jaggies. Can someone explain this?

Not sure why the Sony is not deinterlacing properly. Would either the Lumagen or VP50 fix this?It's a bit hard to say just from a single image, and there are a number of possible variables with gaming systems, but it looks like what I describe as 'jaggies' in my previous post. If the dragon boat is moving fairly slowly, then you might be seeing a degenerate form of combing, where the 2 fields only differ by a pixel or so, but I don't think so (although, again, it's hard to tell).

If it is jaggies (i.e., caused by aliasing) then you'd need a video processor with an edge-adaptive deinterlacer to fix this. The VP50 would do it, as would a VP30 with ABT102 card. Processors with HQV (e.g., Vantage) or VXP (e.g., Crystalio II) chips would likely help, as would one with an older DCDi chip (but not the newer Genesis chips which have significantly degraded edge processing). You wouldn't want to use something with a SiI504 deinterlacer like the Lumagen, a VP30 without the ABT102 card, or an older iScan.

Your best bet, of course, is to try this with one or more of the possible video processors to see if it actually helps. That can be difficult if no one in your immediate area carries these. You might try posting in one of the gaming forums here to see if anyone else has seen and (hopefully) solved this problem.

- Dale Adams

Thanks for your thoughts, Mark and Dale. I didn't mean to hijack this thread.

I don't think it's combing where motion causes the odd and even fields to not line up. The edges never show a saw-tooth type pattern, only blocky stair steps.

FWIW, I did figure out how to set the GameCube to output a progressive signal. My TV confirms an input of 480p--but the image looks the same! Same blocky stair steps on the diagonals. So maybe it's not a deinterlacing problem, but a scaling issue. But then why would some areas be scaled nicely and others poorly?

I don't know what this all means. I think the 70XBR2 has a reasonably good scaler and deinterlacer as SD video programs and DVD film look very good. Is it something inherent in video games? Would any VP be able to fix this?

Hi,

I have a Gamecube (GC) connected to my set by going out via component cables into my Crystalio II and then into my DLP set. I dont see any problems with the gamecube image.

Are you aware that only the EARLIEST Gamecubes were designed to output 480p. There was such little damand (according to Nintendo) that they changed the design of later Gamecubes. You need a special Nintendo Component cable to get 480p on these early GC's along with a special digital output port on your G's. I don't remember if you have to go into the setup menu and change something there as well.

Yep, I have an early GameCube, one with a digital AV out port connected to my display with component cables. If you hold down the "B" button when turning on the GC, you get a menu choice of progressive or interlaced output. Picture looks the same whether 480i or 480p is selected.

So you see no jagged diagonals as in my photo above? The Crystalio II is nice, I wonder if the less expensive Lumagen would do as well.

If you are outputting 480i from your Wii, then the Wii is not doing any deinterlacing. If you are outputting 480p from it, then the wii is doing deinterlacing. Since you have the atifacts with both 480i and 480p out, I would concur that the problem is caused by aliasing.with a cure being a procesor with edge adaptive 480i deinterlacing. The Lumagen uses the Sil504 or 480i video deinterlacing which I understand not to have adge adaptive deinterlacing.

Well- here's another plus for the CII (hopefully Calibre and DVDO are listening)-- it is an order of magnitude easier to upgrade firmware than on the VPs. I've had a DVDO unit which requires purchasing a special cable and installing software on a computer and running this program, etc., etc. I've also had a Calibre Vantage-HD to which I also had to run a wire from computer to the unit, and reset the unit to factory specs and retype my settings, etc.

Well for the CII, I downloaded the firmware and put in on a USB datakey (that they provided with the CII), stuck the key in the USB slot on the CII, and selected from the menu- update firmware. 30 minutes later the firmware had completed the update, the unit rebooted automatically, and all my previous settings were intact.

Can someone explain why the DVDO and Calibre units have to be such a pain-in-the-a-- to upgrade? Why can't they also use a usb port (which by the way is how easy it is to upgrade the firmware on my HP MD6580 DLP tv as well)????

Well- here's another plus for the CII (hopefully Calibre and DVDO are listening)-- it is an order of magnitude easier to upgrade firmware than on the VPs. I've had a DVDO unit which requires purchasing a special cable and installing software on a computer and running this program, etc., etc. I've also had a Calibre Vantage-HD to which I also had to run a wire from computer to the unit, and reset the unit to factory specs and retype my settings, etc.

Well for the CII, I downloaded the firmware and put in on a USB datakey (that they provided with the CII), stuck the key in the USB slot on the CII, and selected from the menu- update firmware. 30 minutes later the firmware had completed the update, the unit rebooted automatically, and all my previous settings were intact.

Lumagen is even easier - I just hook up the supplied comm cable; run the exe FW update file; press a button and the firmware will be updated in <5 minutes with all settings intact. I do have to run the FW updates twice occassionally when the updating died half-way for unknown reasons. Kind of used to it - all Windows applications died like that.

If you are outputting 480i from your Wii, then the Wii is not doing any deinterlacing. If you are outputting 480p from it, then the wii is doing deinterlacing.

Actually, when the Wii is outputting 480p from a game it's not deinterlacing, it is actually generating 60 unique full resolution frames per second. It does this when it's outputting 480i except that it throws away half a frame vertically to create a field, and interlaces it for output.

OK. So if its 480p, is there still aliasing?

OK. So if its 480p, is there still aliasing?

Yes, it's just the way the graphics are drawn. There isn't much anti-aliasing in the graphics hardware, so when it's scaled, you get bigger jaggies. Different scaling implementations may help smooth it or not. This may be one case where you don't want a super sharp result.

You're making me tempted to buy a component cable for my gamecube and drag it down from the loft to do a test through CII.
However, I currently agree that this is in the source - GC is actually one of the better consoles for antialiasing, but it's still there.
I've never had it running 480p into my system though because I was using an HTPC with Sweetspot until recently. To make life easier I just ran all my consoles S-Video.

If I see jaggies as defined as that on Wii, I'll be pretty disappointed.

However, I currently agree that this is in the source - GC is actually one of the better consoles for antialiasing, but it's still there.I'm still confused about GameCube scaling. If you look back at the photo I posted here (http://www.avsforum.com/avs-vb/showthread.php?p=9021798&&#post9021798), the large letters "Zel" and smaller "THE LEG" are very nicely scaled with good anti-aliasing--no blockiness. The animated "cartoon" part of the image, however, appears to have no advanced scaling algorithm applied at all!

X
X X
X X X

is clumsily scaled to this:

X X
X X
X X X X
X X X X
X X X X X X
X X X X X X

instead of taking advantage of 1080p resolution to produce this:

X
X X
X X X
X X X X
X X X X X
X X X X X X

How or why are only parts of the image scaled properly? If my display is not doing what appears to be simple scaling, why would a VP do any better?

This is my take on this.

There's no scaling going on at all. Gamecube is rendering 480p.
Nintendo (as is usual for them) are using their in-game 3D engine to create the ship behind the lettering. This suffers from their 3D rendering algorithms lack of perfect anti-aliasing.
However the logo is a static transparent graphic and isn't rendered using the 3D engine so is created in photoshop with no aliasing present.

IMO a VP wouldn't do any better. But it would be nice to get confirmation - hence why I was tempted to dig all my stuff out.

This is my take on this.

There's no scaling going on at all. Gamecube is rendering 480p.
Nintendo (as is usual for them) are using their in-game 3D engine to create the ship behind the lettering. This suffers from their 3D rendering algorithms lack of perfect anti-aliasing.
However the logo is a static transparent graphic and isn't rendered using the 3D engine so is created in photoshop with no aliasing present.

IMO a VP wouldn't do any better. But it would be nice to get confirmation - hence why I was tempted to dig all my stuff out.

I have a Gamecube connected to my CII via component cable. Do you want to suggest any tests for me to run with any specific games?

yep - just do exactly what Blue 911 has done and see if the jaggies are still present.
So you need Wind Waker launch screen as above, 480P output through component.

Just for fun you could see if there is any difference between 480i and 480p as Blue 911 reported no difference which is the only confusing thing for me.

If you're going 480i at any point it might also be fun to try the Faroudja deinterlacing to see if there is any difference

This is my take on this.

There's no scaling going on at all. Gamecube is rendering 480p.
Nintendo (as is usual for them) are using their in-game 3D engine to create the ship behind the lettering. This suffers from their 3D rendering algorithms lack of perfect anti-aliasing.
However the logo is a static transparent graphic and isn't rendered using the 3D engine so is created in photoshop with no aliasing present.

IMO a VP wouldn't do any better. But it would be nice to get confirmation - hence why I was tempted to dig all my stuff out.You're on to something. I think I've figured it out!

I was assuming that the stair steps were made up of individual 480p size pixels which were not being scaled very well when enlarged to a 70" screen. What's actually happening is that the GC's graphics are so crude that each stair step is made up of multiple pixels so that the scalar does not recongize these curves or diagonals as smooth lines. The scaler thinks that this stair-stepping is the actual shape of the image, so the stair-step edges are faithfully scaled up.

Difficult to explain clearly, but does this make sense? You're right, in this case, no scaler will smooth out these jaggies!

This may have been mentioned above - but do any VPs perform work on diagonals which are already full progressive frames? I was assuming not - so it wouldn't actually matter if the stairs were 1 pixel or bigger.
One reason why 480i and 480p don't look different is that perhaps that it's running at 30 frames per second - so the display is only having to do a simple 2:2 detection to recreate the original progressive frames.

I just connected my Wii with components to my Sammy 3241 LCD, and with the Wii set to 16:9 and 480p, I get scan-lines/aliasing galore. It goes away if I set the Wii to 480i.

Is this consistent with what you guys are experiencing? I'm hoping it's not my cables, which are third-party from Ebay. Also, my Pioneer DVD Player set to 480p looked very nice on the display, and the PS2 looked better as well using components. I really hope it's just the source, and not the displays handling of the signal.

This may have been mentioned above - but do any VPs perform work on diagonals which are already full progressive frames?
I don't think so. I believe diagonal edge interpolation is just a de-interlacing technique, not a scaling process.

Hell of a thread though. I was hoping to put my Vantage upagainst a Gennum processor using an HD_DVD source and 1209S PJ.

unfortunately my wife found a mouse in teh kitchen cupboard and considered it to be an emergency :( I thought it could have waited until tomorrow!

Nick

I just connected my Wii with components to my Sammy 3241 LCD, and with the Wii set to 16:9 and 480p, I get scan-lines/aliasing galore. It goes away if I set the Wii to 480i.

Is this consistent with what you guys are experiencing? I'm hoping it's not my cables, which are third-party from Ebay. Also, my Pioneer DVD Player set to 480p looked very nice on the display, and the PS2 looked better as well using components. I really hope it's just the source, and not the displays handling of the signal.
That's really odd. The 480p signal from the Wii ought to be easier for the Sammy to handle.

And Blue911, I use a Gamecube through s-video (through my vp50) at 480i and I also get pretty strong aliasing/jaggies*. They obviously antialiased the diagonals on the Zelda logo "manually" (although I'd be interested to know how that works as it's against a varying background colour as the boat moves behind it, maybe osme alpha channel thing??) whereas there's no AA on the 3d scene. It looks a LOT better, even on the low latency Game Mode, than going direct to the TV though. I've got a hastily soldered component cable somewhere, I'll see if I can dig it out and try 480p out but I'd expect it to be very similar albeit with tighter colour.

* I think it's down to the game too. Mario Kart is very smooth. I think Zelda's art style lends itself to bold, thick black outlines, hence the effect you're seeing. I wouldn't call the GC crude though, some fantastic looking games on that little machine.

OK guys. Enough with this gaming stuff. Don't mind a little and I am not at all upset. Just let's move the discussion to another thread.

Thanks. Agreed.

Apologies.

No apoligies needed. We are all here to help each other. My concern is that the gaming stuff you guys are posting will be lost later and won't be found by others that might need it. A seperate thread will draw the right people in and make it easier to find in the future..

Tell us more about your VP comparisons. You were last comparing scaling capabilities. How about 1080i deinterlacing? Any differences between various VP's?

Interesting thread. Seems like the upcoming Lumagen Radiance w. the VXP processor for $3500 is going to perform well if it does what it claims here:

http://www.gennum.com/ip/pdffiles/vxp_brochure.pdf

Will see the Lumagen Radiance in 2008 :eek: :eek:

Will see the Lumagen Radiance in 2008 :eek: :eek:

Is this really the level of intelligence, and helpfulness, you want to be known for on this site?

BTW: We will be showing an Alpha level RadianceXD at CES booth South 21666. We plan first Beta ship at the end of January for a small number of units and more at the end of February. Certainly, as you so kindly point out, these dates could slip a bit.

I am sure the results will be worth waiting for.

Jim,

Any particular reason not to support SDI-HD-SDI in the Radiance? Will that be possible as an add on?

Just curious,
Thanks for the information
Chris

As I understand it, there will be an external SDI to HDMI plug in ala Algolith but with the ability to pass blacker than black. From what I nderstand, HD SDI will not be an option. Having the ability to input SDI through one of the HDMI inputs will protect the many of us all of us who have SDI out modified DVD players. These players usually don't have HDMI outs. There is little consumer stuff out there which has been modified to output HD-SDI.

Any particular reason not to support SDI-HD-SDI in the Radiance? Will that be possible as an add on?

We plan a SD-SDI to HDMI converter external module.

We don't think we will do a HD-SDI to HDMI converter, but there are some out there that should be able to be used.

[QUOTE=jrp]Is this really the level of intelligence, and helpfulness, you want to be known for on this site?

Intelligence thank you for letting me read yours :p :p

I see...and no loss from that conversion SD-HDI to HDMI?

[QUOTE=jrp]Is this really the level of intelligence, and helpfulness, you want to be known for on this site?

Intelligence thank you for letting me read yours :p :p

Jim-- please ignore the NOISE in this newsgroup (e.g. MADDOG). We all appreciate the insight into your forthcoming products and with any alpha/beta product its impossible to fix a firm release date. The fact that you are offering to members of this forum the opportunity to purchase one immediately and sign an NDA to be a beta tester is an outstanding opportunity. If I hadn't just purchased the CII, I would have taken you up on that offer.

OK. Back to the subject of the thread.

Got some news Mark?

Not yet. Christmas is a very busy time for a retailer like me. aAso my wife is away for a few days. Thus, the care of the large dog, the cat, the McCaw,and the fish are up to me. The big dog really doesn't like being in the HT. Also over the next 16 days, there will be 10 days and/or nights of NFL games. Most games are broadcast in 720p and all I will be using is the Lumagen because of its superior scaling. I will get to some 1080i deinterlacing testing soon.

I see...and no loss from that conversion SD-HDI to HDMI?

An HD-SDI (or SD-SDI) can, if done correctly, convert to HDMI 4:2:2 (any rev of HDMI has this), without loss. The SDI is 10-bit 4:2:2 and the HDMI 4:2:2 is 12-bits.

Since we are not planning the HD-SDI to HDMI version you would need to check with the appropriate vendor.

For SD-SDI, our convertor will be bit accurate in the conversion to HDMI 4:2:2. If the output goes to a DVI device, it would revert to 8 bit RGB, and so would not be bit-accurate. It would still translate very well of course.

Mark,
Based on your tests to far, which product will do the best job on dvd movies? (non HD and Non Blue Ray, just the plain ole SD stuff)
thanks,
kadiddle

Thank You Jim for the information.

An HD-SDI (or SD-SDI) can, if done correctly, convert to HDMI 4:2:2 (any rev of HDMI has this), without loss. The SDI is 10-bit 4:2:2 and the HDMI 4:2:2 is 12-bits.

Since we are not planning the HD-SDI to HDMI version you would need to check with the appropriate vendor.

For SD-SDI, our convertor will be bit accurate in the conversion to HDMI 4:2:2. If the output goes to a DVI device, it would revert to 8 bit RGB, and so would not be bit-accurate. It would still translate very well of course.

On film DVD, I prefer the Lumagen because of its clean, non ringing scaling. Everything I have seen deinterlacers SD film very well. The dinterlacing improvements of the new stuff are mostly with respect to video. With 1080p 24fps displays, HD film will go straight in. 480i and 1080i video are where the deinterlacing improvements lie for the most part. Mosquito noise reduction, other noise reduction, detail enhancement, these are features ofthe new stuff over the old.

It's 3 AM and I hear noise up on my roof. Actually, I am staying up not to catch a glimpse of Santa, but because I have to drive my daughter to the airport at 4:30AM. She and her boy friend are off to Mexico.

Thanks Mark,
Is there a link where I could get explainations of the terminology you use, such as Mosquto noise etc etc?
kadiddle

Hi,
I am considering buying an external scaler and would like some advise.

First of all, is this the right thread to post in?

My primary use would be to make SD DVDs look as good as possible, but I will also have VHS and SD satellite input, and maybe HDDVD. My display is a panasonic plasma 50ph9uk (768 line native).

I have been considering the LumagenHDP so far, or perhaps one of the DVDOs.

I will probably use an oppo 970 dvd player as the source since it offers 480i over HDMI, but I have also been trying out a Toshiba A2.

The A2 does a pretty good up conversion of SD as it is, but I am wondering if the 970/Lumagen VisionHDP would be even better? The A2 has bad jaggies, but I hear the HDP is no better in this regard. I do like all the control the HDP offers. Is there anything worth buying for around 1000 bucks?

Thanks for any info.

-RIch

Rich: You shouldn't see "bad jaggies" on DVD through a Lumagen, unless all you watch is music video's...and even then I don't think they would be that bad. If you watch primarily film source dvd then it's going to be great.

Gordon

Rich: You shouldn't see "bad jaggies" on DVD through a Lumagen, unless all you watch is music video's...and even then I don't think they would be that bad. If you watch primarily film source dvd then it's going to be great.

Gordon

THanks Gordon. Now I understand. The "jaggies" tests on HQV must be recorded from video so there is motion between fields.

What about the video delay? I hear the Lumagen HDP will introduce about 50ms delay. WHy don't they include an audio path through these units as well (as I see Genfen is doing)? THen we could use only one single switch for both A&V and delay the audio to match the video in one neat unit.

My receiver (sony ddw900) has only one delay option, 68ms.

WHat have you heard about the soon-to-be-released Genfen's? THey seem to promise an awful lot for the price.

Thanks for you help,
Rich