New DVDO iScan Duo [2.0 firmware released] - Page 106

Exactly! Some were told to never touch the offsets back when the RS1 came out but what happens when the bulb ages the lumen across the visible light spectrum drops in red and green but blue doesn't degrade as quickly. So the image color temp goes up. Drop blue and sometimes some green to get the grey scale as level as possible first. Then run auto Cal. I got great numbers last night. As long as its under DE of 2 you can't see the difference and auto Cal is so easy just do it more often. It took so long to do manual adjustments I would put it off longer and end up with DE's of 10 or more. Auto Cal is awesome!

I suppose if you're a nitpicker then do it manually but no need to really IMHO! With a projector the colors shift quickly anyway because of the bulb. Heck I lived with the Lumagen HDQ CMS or what Lumagen called a gamut mitigation tool for two years and was happy with that so a DE of under 2 is fine with me. LOL

Can anyone tell what sort of scaling algorithm the Duo uses? I'm hoping for something that interpolates pixels in a way that approximates normal light mixing (0 black and 100% white = 72% grey if source is 2.2 gamma encoded, not 50% grey as most image scalers I've dealt with would come up with.

Put another way, does it interpolate the already logarithmically encoded luminance values using logarithmic math, or does it make the usual mistake of using linear interpolation on the logarithmically encoded luminance?

I know it seems like a bizarre thing to ask; if such an expensive piece of gear applies the right kind of math to the problem, but I see it done "wrong" so much, that I wouldn't be at all surprised to see linear interpolation used in something like the Duo. Apparently, linear interpolation is "good enough" for most, and is a cheap way to increase acutance (the effect on edges is a bit like a sharpen filter), but I don't like it, because it causes fine textures to appear too dark.

I can show some examples of "good" and "bad" scaling on request if needed, to demonstrate what I'm talking about.

Please do!

[edit]Huge images removed, they have served their purpose.[/edit]

Hrm.. If interpolation can be disabled, all my concern is 100% moot.

On your so called "good scaling" shots, the plain grey areas are altered in their luminance, while the luminance remains true to the original in your "bad scaling" ones.

Otherwise, quite interesting. But how important is this in real-life performance ? This would imply that gamma and greyscale should be set with checkerboard patterns instead of plain grey fields, right ?

No, the plain grey stays true to the original in both linear and logarithmic interpolation, it's the dither areas that appear darker with linear interpolation. The effect is that the darker pixels get too much weight against the lighter. If the solid grey looks changed to you, that's an illusion, you can drop them into an image editor and use the eyedrop tool to check

Anyways, the real life difference is that fine textures come out too dark when linear interpolation is used on a logarithmic image. It appears grungy to me, it's as though textured areas have a lower gamma than plain.

That XRGB-3 looks really nice for 240p, but not so good for 480i, where the duo might be better.... maybe I'll just get one of each XD

I HATE the ringing from the VP-50pro, so if that's not better in the Duo, it's a big issue to me. Duo doesn't mention two different game modes either, so there's obviously some differences.

The reason the gamma charts use both a dither of black and white and a grey is that black and white stay the same during gamma adjusting, and the dither pattern blends optically in the eye when seen from enough distance to make a grey reference. The actual grey, the solid areas is what changes when the gamma curve is adjusted. It's a clever test pattern for setting up gamma quick and dirty without a luminance gauge.

The first one with lines going different ways and with checkerboard too serves not only as a quick gamma check, but also as a test for ringing/soften/sharpen type distortions. If a screen is setup perfectly, then that pattern would appear as nothing but a big grey rectangle from a distance, with the exception of the white and black bars in the middle.

Such are only recommended to be used at native resolution for adjustments, because the interpolation during scaling tends to distort the dithered parts, which is what I demonstrated, and was wondering how the Duo does interpolation, whether it would take, for instance, a pixel of 100% luminance (255 rgb), and another of 0% (0 rbg), and arrive at 50% (185 rgb) between them, or if it would do linear interpolation on the digital rbg values and arrive, incorrectly, at an rbg value of 127 (22% luminance) for the same pixel, as very, very many scaling algorithms would. Not because I want to use upscaled gamma patterns for any adjustments, but simply because I'd prefer it if when deriving the intermediate color between pixels, that the scaler would do it based on the actual luminance line between the pixels, the way light actually blends in the real world, not the digital color value, which is logarithmically (base 2.2 normally) proportional to luminance, not linear.

I know I'm rather picky, but these things aren't cheap enough for me to buy lots and just use whichever I like best, so I'd like to find out the peculiarities of them before buying stuff that (usually) can't be refunded.

With scalers that don't interpolate, none of this matters, but I didn't even realize there were any that didn't interpolate (such as the X-RGBs and several others from the looks of the screenshots.)

This is starting to get OT from the iScan Duo now though...

I did, 120/120/120 vs 116/116/116 on your top most screenshots - the large grey area in the lower left corner


The XRGB-3 is outdated. Get the XRGB-Mini. It's *ACE* for 240p and 480i.


the scaling engine is practically the same, so the Duo's not for you.

Nevertheless, very interesting thoughts on scaling engines affecting the luminance of different detail levels differently!

That difference was an oopsie somewhere in the conversion to and from linear luminance, wasn't supposed to happen. Probably inadequate precision in my image editor.

Anyways, I'm probably going to end up with a Framemeister as per your recommendation, it looks really nice to me (and cheaper to boot)

My Duo's back!!!

That's good news.

Well after many many many many years of having DVDO products I've gone over to the dark side. . Just installed a Lumagen XE3D into the rack. Might keep the Duo for another part of the house or sell it..... SJ

Congratulations SJHT on your purchase. If you ever get a chance would love to read the comparison between the two.

Will do. I can already tell this thing screams "quality". SJ

Looking forward to it.

Question, I had my Mitsubishi WD82837 calibrated by Chad B. using the CMS in my Duo. Does anyone know if I will lose my calibration settings in the DUO if I upgrade the firmware?

I upgraded from 2.30 to 2.32 and did not lose my calibration.

Me too! Is the XE as sharp with bluray or softer looking?

There is a review/comparison of sorts in May's Home Theater magazine between the Duo and the Lumagen Radiance, if that's of interest.

Thanks prepress.

Yeah, I read that review prior to my purchase. I agree with the reviewer that the Duo really is the best value. SJ

Finally did an update, hopefully all goes well for Thursday for calibrations.

An update? I don't see any updates posted on DVDO's site. Am I missing something?

There is no new update sorry, I've never updated from 2.20 and this is the first time I'm using version 2:35 and luckily I found someone with calman experience in my area so it will interesting to see how it works with the duo.

Btw when it connects via rs232 does the computer control it from its end?

Thank you.

2.35? You mean 2.32!

Yes, assuming you will use a RS232 to USB connector to your laptop, it will be controlled by CALMAN.

You are correct.

Thanks fight4yu, how long are those cables usually?

I think you can get the in all different sizes. The ones I have is 3', I believe. It's longer than the cable on my Display 3, which is the limiting factor to how far away I can sit when calibrating.