[wimver]

I am rendering on both platforms, with camera and tone mapping settings identical. yet I get different results. should I change the monitor gamma of either system? or what is needed to get even results?

tnx
wim

[mverta]

Are both systems being viewed through identical monitors, or at least monitors capable of displaying the same colorspace, plus hardware calibrated, and properly colorspace profiled?



_Mike

[wimver]

no they are not, I was not aware that MW is depending on external hardware (monitor) calibration to produce images.
I do not depend on monitors to compare the images, I render images with identical background on both systems and then import them in photoshop for comparison. my monitors can as well be off in the whole process. since there is no color profile attached to MW output files, (shown while importing in PSD) I presumed they had generic RGB values and hence should be device independent...
what do you suggest in terms of calibration software? or how can both systems be matched?

tnx
wim

[mverta]

Maxwell doesn't require calibration to produce images any more than any other software does. What I'm saying is that if your monitors are not both equally capable, and equally calibrated to profiles, you can't make comparisons between them anyway that mean anything. Otherwise, the whole thing is arbitrary: which one is right? Colorspace profiling can make a huge difference, especially in Photoshop (the OS's themselves all run in sRGB), but before you even deal with that, you have to start with the monitors themselves, and what's driving them. If you preview an image in something other than Photoshop - something which doesn't do colorspace profiling, like Quicktime Viewer or some other Windows utility and they don't match, then at the monitor level, the monitors themselves don't match. You have to start there.


_Mike

[wimver]

No Mike, I think I did not make myself clear. I render image A on a mac, and image B (almost identical to A except for 1 detail) on windows. in Studio, all settings are equal on both platforms. monitor gamma is set to 1.8 on both systems.
after renders are finished, I open both images in photoshop on the mac. The images should be identical, but they are not. the image coming from the PC is much more saturated.
and yes, I tried with gamma 1.8 on mac and 2.2 on windows, but still did not get the same result. I tried with a variety of gamma settings, I just can't match the 2 images.
I did a test on another mac (laptop, so different monitor from the eizo) and got identical results to the first mac.

I want to know which setting on windows should be different from mac settings.

thanks
wim

Ps. I have worked for Barco for 15 years, so I know a little bit about monitor calibration

[mverta]

Forgetting what they look like, have you sampled identical pixels in each image to see that in fact they're different values? Before you bring the PC image to the Mac, I'd also sample a pixel in Photoshop on that system to see its value, and check the corresponding pixel value on the Mac version. This is assuming, again, that the scenefiles, materials, texture maps, everything are identical.


_Mike

[wimver]

Ok, I did a new test to clear things out.
here is how:
1. save "pack & go" scene
2. open this scene in studio on both platforms
3. set correct output paths (directed to a folder on the mac)
4. render
5. open both images on the mac in photoshop

I get different import messages. the mac version has generic RGB embedded, the win version has no embedded profile.
how do I put a generic profile in the render?

[tom]

OK, now the problem is clearer. You should know any image rendered using 1.7 (under any platform) in fact uses sRGB color space but it's not embedded in the image files. So, PC PS says it has no profile and Mac PS says it has a generic profile. At this point, your decision is heavily critical and have only one correct answer, it's sRGB. Even, if your workspace is something else than sRGB and you must assign sRGB to the render you're opening. Then you can convert it to working profile (eg. Adobe98) or keep working with it without converting it (an asterisk will appear in the title of image). Anything else will damage the output.

[wimver]

Aha, now this is better! they match!
thank you so much Tom!
rgds
wim

[wimver]

But why exactly sRGB? this is such a limited colorspace... After all, the original is 16 bit, so isn't that heavily overreduced?
I still remember my first lesson in color calibration, where the tutor almost shouted "Never, EVER use sRGB... it is yet another proof of microsoft's bad taste..." (... unless you only plan to produce web images...)

wim

[tom]

sRGB is not related being 8 or 16 bits or more. That's the precision Maxwell renders in full visible gamut but fits the output into sRGB space because 99.9% of today's displays is only capable of displaying sRGB.

[JorisMX]

tom, when saving a hdr rendering after decreasing the bitw from 32 to 8 does this mean the same as for pngs and tifs?

I prefer working with .hdr output due to banding issues etc

[tom]

Yes, sRGB is still valid in that case.

[JorisMX]

Thanks Tom,

As I do cgi for print most of the time I would LOVE it if there was a tab or extra function for saving to adobe or even better eci-rgbV2

I don't really understand why NL is choosing to reduce themselves or rather maxwells output to a colorspace used for web only.
Wimver is perfectly right, from all commonly used colorspaces sRGB is the smallest.

And yes, people that work on office computers/non-workstations may have sRGB profiles still assigned to their screen.
But we are using proper workstations, calibrated monitors etc.
And you can always do a little postwork and even color-profile matching by using softproof in PS to make sure your rendering will look pretty much the same on a standard screen.

I mean when I mic a guitar amp for a nice production I will record to 96Hz 24 bit and then (finally) convert and dither down to the 44.1/16 bit we all know and hear everywhere. No reason to limit yourself from the start.

[wimver]

sRGB is not related being 8 or 16 bits or more. That's the precision Maxwell renders in full visible gamut but fits the output into sRGB space because 99.9% of today's displays is only capable of displaying sRGB.

I am sorry to say so Tom, but this really does not sound clever.
first, I know that srgb is not related to 8 or 16 bits, but out of 16 bits any colorspace can be distilled, and reducing an image to the smallest possible colorspace does not leave much space for finetuning, let alone improvement.
and second, as JorisMX points out, they do exist, people who render for magazine quality publishing.
I haven't had any complaints so far, but there is good and there is better.

wkr
wim