Cineversity

I have encountered this problem many times and I cannot see how to solve it…

When I render in RGB I get the most brilliant colors. But when I open a still or rendered Quicktime Movie it is bleached to a disappointing point.

How can I fix this?

Hi Albert,
Would you mind sharing the scene file and a image as well the Quicktime file? Perhaps via Dropbox? I can set up an upload link, if needed.

Color profiles and color management comes first to my mind. I exclude here the “auto-gamma” setting from Quicktime, as you have the same problem with images. Will this happen witha psd 16bit/channel file as well?

Anyway, guess-work is not really efficient and takes only time away from you. Please let me know what kind of screen calibration you use, how it looks like if you do a fresh calibration or how it looks like if that option is switched off, as this is nothing I can see here of course.

Since when do you noticed this?

Perhaps it is a case for the support, but I’ happy to look into it.

All the best

Hi Sassi,

I noticed it, I believe, ever since upgrading to R16… or even R15. There are distinct color differences.

Since you can see both ‘color versions’ in one screenshot next to each other there is definitely a mismatch somewhere that is not consistent between apps…

If you have a dropbox link I could send you the files…

Albert

Thanks for the files, Albert.

The core of the problem starts with the “EIZO Photography Color Profile”. This is set in you render settings.

This is normally not wrong, but if any step after that ignores color management, you have a problem. Ignore color management, means for example that an sRGB profile is expected, as mostly in JPEG.

If I simulate this in Photoshop, I go to Edit> Assign Profile …

In the moment I set this to sRGB (with Preview on), then the colors fade or bleach to take the term you used.

NOTE: Assign means to IGNORE any Color Management that was done before. A dangerous option if the previous colorspace is not known.
The option to convert takes the pixel information, calculates the values based on the profile and then moves it into the colorspace of the new one. Whit a lot of options to mess up things, e.g., if the target Color Gamut is smaller colors can shift or even clipped.

To just render in any Color Space, without even considering the pipeline after is a quality loss. For Quicktime movies that would be REC709 for HD and for UHD REC2020, the closest (not matching!) to REC 709 is sRGB, but not the same. Certainly not any Monitor profile. That might work locally but I would suggest to check the target and its needs.

For any images with critical color values and the idea to convert those in a next step, anything 8bit/channel and even worse, have dithering enabled (8b/c) will lead to low quality. I suggest since long to work at least in 16bit/channel, and even if the delivery is 8bit/channel the pipeline for anything quality can’t be 8bit/channel.

If you like to work that way with a specific color profile, you have to know what comes after and how the file is “interpreted”. Yes, a decade ago, I would have answered, it is just RGB, you have to deal with it in post. Those days are gratefully over, and yes, that is the change you noticed.

I hope that helped

I will soon add an one minute clip!

Hello Sassi,

Thank you for your explanation. I did the EIZO (my monitor) as a last grasp guess to solve the problem. Before that it was set to sRGB… I have included three files and the other two should still be set to that original format?

Is there anyway to load a simple preset render settings to just have the same color I see in the picture viewer AND in the finder on Mac? If I save out a series of stills in tiff in the picture viewer I see rich saturated colors but in the finder they already look markedly different. And they have not left my machine yet?

Albert

Hi Albert,

I saw the one file with this profile and saw the problem. The other one is clearly, avoid 8bit/channel, especially since this dithers the pixels to look kind of OK. This is a big no in my book, but yes, all too often I see this as production quality, which is no, not even close.

The color profiles used change the values in the file, the profiles stay the same. So with 256 values per channel, you are already at the lowest end possible (excluding some lower binaries at all of course).

The dithering tries to mix up values by lifting and lowering values of different pixels to SIMULATE and make believable some color values. Works for the web as delivery, more or less. BUT in the moment you change the profile, these artificial values from the dithering are seen as what they are and not longer in concert. They become again moved, and when saved, perhaps dithered again. To even think about color-fidelty then is not even possible.

The color profile sRGB is the smallest color gamut in use, from where I look at it, it is for delivery, and not at all for production. Since everything is normally in Linear and floating point, sRGB is not so much of an concern anymore, during production. But 8bit/channel excludes that idea of course from the start.

What you see on your monitor is based on many parts in the whole pipeline. Considering all is calibrated, from the texture generation (real world) to your scanner, and monitor, all the way to the delivery (in which format is based on the facility of course). There is more and a really long story, like some people do not even give you the Color temperature when they photographed the textures, so you might hit it or not at all. Weird stuff to pay even for that.

What ever you see on your monitor is based on many parts, even in Photoshop, did you switch on Proof or not, do you use sRGB or Monitor RGB, it changes the content. If an image has an ProPhoto color profile you might not see even all the colors, compared to an sRGB image. Again, this theme fills books and I’m certain that I can’t give you here more than a hint.

In the clip that I promised above, you can see the changes clearly. What method one uses to convert is also certainly based on many things, is it for a short time and you stay in a large space (app based) or is it for the web (delivery) will make a huge change. Often the Perceptual is comfortable, and I prefer it for many areas, but it can be also limit the result, if one has to go back and for.

https://www.amazon.com/clouddrive/share/3tj0tdl7Qwiev8QjGZr2x6ZwjNo2kFZFxsv4LTF3MuC?ref_=cd_ph_share_link_copy

So, to your question, I render the other scenes even missing a texture in one, then saved it and reloaded it, as in sRGB, with an A|B comparison, I used (while holding down the CMD key in Picture Viewer) the eyedropper and get exactly the same values for the color. So if you see something different it is a question of color profiles. With Quicktime, it might be a this and/or the Auto Gamma settings several codecs offer, e.g. ProRes HQ.

I can’t see the difference here in the Picture Viewer.

I can’t tell more, my best suggestion would be then: support.
http://www.maxon.net/support/support-questions.html

All the best

Hi Sassi,

Unfortunately I don’t understand all you try to explain (my fault…)
When I load the image back into oicture viewer I get the exact same colors. But when I view one outside of cinema (in the Mac OS finder window) and the other in picture viewer I see the different colors of the screenshot. So that probably means that the display colors differ from the colors in Cinema’s windows…

I also don’t understand why 8bit dithering is standard checked in Cinema (I never touched that…) if it is so bad.

Thanks still for your time and explanations!!

Albert

By the way, these files never go to professional video… they are mainly for clients and their (regular) PC’s…

Hi Albert,

I’m certain that my text, even not short, is not an approriate representation of the subject considering all aspects and details. It took me a while to get a clear picture of all of that myself, considering all the misinformation available on the web. I took finally three Color Science courses each three month long with Dr. Poynton, e.g., co-author of many SMPTE material, besides a lot of other material, to cut though diffuse information available.

Please check all in Photoshop, with the approriate settings. I’m not certain what the preview does, nor do I have found a white paper about it.

The dithering is not bad by itself, as it has helped the 8bit.ch ;o) to survive for that long time. Without, you would see “bandings” on many areas where gradients are.

BUT, if that file is altered in any way, things become maybe more ugly.

I have no understanding to use this bit depth in anything other than preview. Even the TV stations didn’t took anything below 10bit/channel since the late ‘90s!

I have attached an image, with the A|B and the Photoshop and Mac Preview. Note that my Mac has a Retina display with a P3 color space Which allows to show more than my previous monitor. It allows me to see differences that I wasn’t able to see on the other, and the idea that the same RGB values should look the same is certainly not a given.
Have a look to the Mac Profile app, and see how different these colorspaces are, just to get an feel how some get squeezed into another space. With a lot of problems along the way, as one can easily imagine when colors are not even available in another space!

https://www.amazon.com/clouddrive/share/uBzjCUiqv4xevnRc0oLa0y4VoledoAZZEWL2aVm8VO2?ref_=cd_ph_share_link_copy

I wouldn’t trust any preview or even the editor rendering. Colorists have highly calibrated screens, in light controlled rooms, with gray painted walls. And yes, if you see it on one screen the idea is, I should see the same. So, how is that working when even Photoshop can set different Proof modes? Long story again. Evenin the image here attached is placed into an JPG with a specific space again and will look different on your screen than on mine, except we would have the same brightness, calibration and screen altogether.

All the best

P.S.:

I have no idea if that works for you or not.

Open the image in Photoshop, set it to 100% in the Navigator: in that way it should be nearly evenly gray.

Then switch the Curve Adjustment on. … or do your own curves…

All I did was to dither the gray value. :o)

Another try, and I’m not certain if that is really the case on your side. But it showcases the same behavior. I created a “rainbow-gradient in Photoshop. Converted it ProPhoto. Since I have done this before and posted such an PRoPhoto based image on my “social Media” I know that this can fail, as the colors of my image were off. Even I included the profile, it was not used and my blue sky look pale purply bluish. Fail So, here I recreate this., but I have not included the profile, to showcase it here.

In preview, since there is no profile, it is assumed to be sRGB and that looks “bleached-out”.

In the moment I open it in Photoshop and Assign the profile, all is fine, save it with the profile and all looks good.

Similar effect, perhaps different “mechanic on your side. But this is my best guess and the most simple one. Something went wrong at one point, but I don’t no how to figure it out what happens. Perhaps check your files (Images) if they work after Assigning (not converting—for this exploration!) different profiles. The one which works best, is the one missing.

https://www.amazon.com/clouddrive/share/sFPAhTlBhruFkg3pfs5atJpC4CFIKtAbd2fznR7H4Iu?ref_=cd_ph_share_link_copy

Ah! Again the watched-your-reply-via-mail syndrome. I did not see your other two replies :-(

Sigh, color seems a really difficult beast when working in both print and RGB worlds…

So I just hope the best for it. Too many apps all doing their thing it seems to me as a lay person.

Happens, and sorry about, Albert, I wish I had a better idea to keep the information (e.g., missing e-mails) flowing.

I remember vividly when I dove into Color management, how confusing it was. Too many self pronounced teachers with incomplete conceptual understanding of the matter. Even in books.

On the end it is simple, but as usual, simplicity is only possible after moving though the complexity. Otherwise it is just a fragmental knowledge.

On top of that and in the mix, the linear light vs gamma discussion, the idea that RGB and CMYK work pretty much in opposite ways (subtractive vs additive) and the current mainstream confusion about LUTs: all of that is certainly not helping.

Simply put, it is a translation system so, each part in the system can produce the same impression, in the boundaries of each. While calibrate, things should appear similar anywhere in the world.

A gap in the chain might not really easily spotted, and that was the discussion above.

My tip, draw a flow chart about your very own visual pipeline; Add conversion points in this flow chart, refine it and discuss it. Try to break it, to see the problems and be able to identify them.

All the best