How to get better VRay renders (aka working in linear space)

[lukx]

I'm still not sure if I get everything right about linear. I set my monitor to 2.2 and set color mapping in vray to gamma correction (bright 0.4545).
Light looks really nice but I have problems with colors. Let's say I want some really red thing in my scene. EWven when I set rgb to total red color is washed out. Maybe I calibrated wrong my monitor. But what about printing it. It's printing also with washed out color.

[thablanch]

How can I avoid using color correct plugin. (can't use it here for some IT considerations, blablabla.. ) is there a reciepie for corection, because if i'm correcting one by eye, i'm loosing the real values of the color

[tricky]

My conclusions:

Argument: using the 'traditional' rendering techniques, where gamma etc are ignored, when I render a scene, my dark areas look VERY dark. I would like to get more light into them. I have a couple of options - add fill lights or play with colour mapping settings a bit. If I add fill lights, placement and setup times increase along with associated render times. Exponential Colour Mapping seems to work nicely, but I find the strong burn-outs from the sun can be lost and the images loose some of their 'oomph'.

Solution A: work in linear space.
This will give me more detail in the shadow areas. Well, I have a piece of callibration hardware on order which will help me properly set up all of our monitors, so thats a good starting point. I need to set Max's Gamma Display to 2.2, and Bitmap Files Input Gamma to 2.2 (so that all bitmap based textures are corrected by default). When I want to use 'colour-based' materials where no texture maps are used, I need to use the free ColourCorrect plugin to make sure the gamma of those colours is also set to 2.2. If I want to work in 'true' linear space (which I understand as a kind of floating point space where possible output devices are not yet known: it may go to a printer, it may go to screen or it may go to DVD...who knows? who cares? Working in floating space will allow me to worry about that later by assigning the appropriate ICC profile in Photoshop), I need to either use Throb's 1.64 curve or the sRGB mode in newer builds of Vray which preview what my render should look like, but will not actually 'burn in' the countering curve to achieve this 'look'. If I go down this route, I need to start thinking about colour profiles and things when I open the rendered images up in Photoshop for post-processing etc. I need to think of this to maintain a workflow in true linear space and ensure that when I send out the final renders they have the correct colour profile for the output device they are being outputted on (eg an Epson 2100 ICC profile if they are printed on my Epson 2100 printer, or a 'blah blah blah' profile if they are to be outputted on a 'blah blah blah' device). However, although very powerful, this method is quite new to me and I haven't quite got my head around it all. So, on to solution B:

Solution B: Work in 'pseudo' linear space.
To do this, I leave the Max Gamma settings at 2.2 (both Display and Bitmap Input) as in Solution A, and I simply set Vrays colour mapping settings to Gamma Correction with Dark Multiplier set to 1 and Bright Multiplier set to 1/gamma of monitor (which will be set once my hardware callibration thingy arrives), which is about 0.455. Now when I render, the 'linearness' is burned into the image so that I still get more detail in darker areas, but don't need to worry about colour profiles and all that in Photoshop - I treat the rendered output just as I have always done.

I really (really) hope I am on the right lines with this. The problem with these forums is that when someone makes a point and somebody else responds, it isn't always clear whose point is being responded to.

[timmatron]

Sounds like you've got it down to me. So far I prefer burning in the gamma, because I think that method a little smaller leap to make than the transition to the floating point format.

[flipside]

Tricky, you confuse some of the terms I think, but basically your way of thinking is correct. I also use your B workflow. This has the extra advantage that for example ir map takes the gamma color mapping into account, resulting in less noise in these dark areas.

Linear means nothing more that gamma 1.0 imo. The floating point issue is that you do your conversions (for example gamma 2.2 textures to 1.0), in floating point instead of in 16 bit or whatever bitdepth that actually discards some of the out of boundary color values.

[fewlo]

Tricky,

After reading your post I finally understand what the f is going on, I too use workflow b, which was eloquently written by your good self.

This explanation has finally made it 'click' (I hope) in this mixed up brain of mine!

Cheers mate, much obliged.

[tricky]

I also use your B workflow

( Cracks open the champagne)

OK, so terminology aside, I think I just about get it. My only question (and I suppose it would force my hand into using Solution A) would be how you supply your final rendered images to the various brochure houses/large format printing companies that you use?

If I was using Solution A, I would simply supply a linear (Gamma = 1) image to the printing company and let them worry about it all . But as I am using Solution B, is it a case of 'send-it-to-the-printing-company-and-hope-for-the-best', or is there anything more reliable?

[tricky]

Tricky,

After reading your post I finally understand what the f is going on, I too use workflow b, which was eloquently written by your good self.

This explanation has finally made it 'click' (I hope) in this mixed up brain of mine!

Cheers mate, much obliged.

No problem whatsoever.

[flipside]

If you do your option A, you'll get a very dark print imo. You need to assign it the correct profile, in this case the linear profile I think.

In the second case, I'm not sure.

But I guess there is more than only gamma to consider when printing. To match your screen color to the print output requires some more calibration of your printer. I guess you need the print company's printer color profiles to apply to your image (or your working space??) so you can see on your screen how it will look like if they print it with their printers.

[dlparisi]

How do you apply the "linear profile".? Whenever I look in photoshop's color settings all I see are sRGB, adobe RGB, etc.

[flipside]

There is a link somewhere in Gijs' tutorial I believe, you need to download and install some profiles in order to access them:
http://www.gijsdezwart.nl/tutorials.php

[tricky]

If you do your option A, you'll get a very dark print imo. You need to assign it the correct profile, in this case the linear profile I think.

In the second case, I'm not sure.

But I guess there is more than only gamma to consider when printing. To match your screen color to the print output requires some more calibration of your printer. I guess you need the print company's printer color profiles to apply to your image (or your working space??) so you can see on your screen how it will look like if they print it with their printers.

OK, lets forget about this for now then. Its beyond my current scope of understanding.

Another couple of points on which I would like your opinions:

1) A colleague of mine has just raised the (valid) point that it is all very well getting more information out of these darker areas, but doesn't all this add up to a less 'oomphy' image, with less overall contrast? Essentially, I suppose it does, but in the end, I think we are dealing with two opposites - that of a high contrast 'oomphy' image and that of a more evenly, subtly lit space. The art of visualisation comes from striking a balance - one man's junk is another man's treasure (was that David Brent who send that?)

2) The next thing is all this EXR business. I have started saving renders as EXR to see what happens. Now when I open the EXRs in Photoshop CS (with the EXR plugin installed), I get further options to modify the exposure and gamma. If I use Solution B, should I leave these both set to 1? I guess so. So the advantage of EXR I suppose is that it gives me a bit more opportunity after the render is done to fine tune things a bit more. Thoughts?

[flipside]

For your first point. Don't forget that with the old workflow, you were adding fill lights all over the place to light up these dark areas, so you were also lowering the overall contrast. Since with this new workflow you start from scratch, and adjust your lighting accordingly, you will also get good contrast, and even more real looking images. You should adjust contrast and balance out your image by adjusting the various lights in your scene, for example skylight vs sunlight vs artificial lights. It needs some time to get rid of the old tricks, but if you get used to the new workflow, your images will look more oomphy than ever.

[armilla]

Nicely summarised tricky! I too am mulling over whether to adopt "Linear Workflow" - probably Gijs' method (your Solution B) - and have yet to wade through the ten pages of this thread. I started to follow it some months back, but gave up when my head started to hurt. The thing is, I don't want to just do it, but would like to understand at least some of the reasoning behind the workflow, and your breakdown has helped with that and makes me want to go back read the thread from the beginning. Just need to find a spare day or so to get my head round it!

What monitor calibrator have you gone for? I definitely want to calibrate my monitors before adopting the workflow, but am not sure which calibrator to go for. They all seem quite pricey, especially for something that I won't use very often.

One of my concerns, is with the Linear Workflow, I could be producing images that look great (well, to my eyes anyway) on my calibrated monitor, but look very washed out, or too dark , or whatever on a client's monitor that is set up with factory defaults. I guess there's no way round that dilema though, you just have to tell them you're right and they're wrong.

Cheers,

Andrew.

[powerandrubber]

I think the beauty in this lies in the fact the you can save your rendered output as always with gamma burned in and at the same time have the vray vfb autosave a vrimg with the "uncorrected" image still in tact available by converting to exr. I basically work with a gamma burned output tga with the comfort of knowing that I can retrieve the raw exr and tweak gamma and exposure in PS where otherwise you'd have to rerender. No need to make a choice because you can have it both ways.