Using digital logos in PBR workflow

Support Materials and Textures
1

I’ve been using PBR materials for most everything these days, which helps when someone wants me to convert their model to real-time, or use it in a different software, etc.
But I haven’t really found a good way to convert graphics that have been given to me and specified an RGB spec, or Spot color.
Great, so this is specified as Pantone 1788c? That has an RGB value, but that’s for on screen viewing. If you use ee263b as the color in your render then it nearly glows.

In general, I don’t just get one color either. It’ll be a logo, so there are several colors.

All the PBR workflow help I’m finding is converting Photos to maps. The only way I could convert these then is to print them out, take a photo, convert the photo, then use that? Well, I could just sample those colors and use them on the original so the graphic is still nice and crisp (they want an odd mix of realism and perfection, so no realistic scuffs, etc).

I’m just curious if there are any tips on how to modify the colors so that it is rendered to look like a photograph of the correctly printed color.

2

Hi, I have the same problem here and I think it’s more a problema related to color management than to the material itself.

I sugest you try change from Filmic to Standard in the Color Management section on the Render Properties tab, but this will make the render not look photorealistic.

Another workaround I make sometimes it’s to correct the colors on post-production.

3

https://www.pantone.com/color-finder/1788-C

4

The color you get in your rendering depends on a lot of factors besides the rgb value in your material. Lighting, Reflections and other things influence the resulting hue, saturation and value. Therefore the only way to get the color right in your final rendering is to tweak it in the compositor.

5

The default Filmic setting is maybe not optimal for display pure tones,since Filmic has a compression around 80% + for high dynamic light renderings.
For pure colors you have to keep your lightstrength within 1.Otherwise you alter or clipping your colors.
And you have to use standart colormanagment.

Like this.In this preview the only light is the backround light with white color at strength of 1.With standart CM settings you can see in the screenshot.

pantone1
pantone1738×1040 52.6 KB

edit,I want to add that in this render, i used a simple diffuse shader, with the hex code as input color.
You could use a emission shader, with lightstrength at 1 too,and black backround.It depends ofc what you want to render.

6

If a colour is specified in both RGB and CMYK then it’s safe to assume the RGB values give a colour which also lies within the gamut of CMYK printing. Just go ahead and use the given RGB values (or hex if it’s easier to pop into Blender).

If the colour values are only given in RGB and they look a bit bright or extreme then it’s worth taking a look at them in Photoshop. Say someone specified a colour 255,0,0 for use in a printed logo. That’s never going to print on a CMYK printer. It’s outside the gamut of CMYK. Putting that value into Photoshop’s foreground colour shows the warning triangle meaning it’s out of gamut for CMYK. Clicking on that warning will change the colour to be within the range CMYK could print. If you had an RGB logo or image with multiple colours, you can hit Ctrl+Y to preview them in CMYK.

I worked in the newspaper industry for 21 years in a production department. We needed everything to go out CMYK, but we got all sorts of stuff handed to us some of which just would not print in CMYK. People that had designed their own logos or the own advert they wanted us just to use. Things like bright vivid greens were the worse case. They will never print in CMYK.

If I have something I want to use in Blender to simulate a printed product then I’d open in in Photoshop. I’d convert it to CMYK. Then I’d convert it back to RGB and use it. In the screenshot here that would mean the bright RGB red would then become a CMYK compatible less bright, less vivid red which could feasibly be printed.

image
image1018×961 65.6 KB

I’d do the exact same thing with spot colours. Typically they might be colours that could be used in a printing process, but only if you have that special ink. Think of thinks like luminous colours. We know these can be printed… just not on a CMYK print process. You need a special ink made up only of that colour/pigment.

7

So I don’t think you many of you folks are understanding the issue. I am not after setting an rgb value and rendering that exact value. What I’m wanting to make sure is that I am providing the correct RGB value, so that cycles, eevee, or realistically Any render engine treats it properly according to it’s physical values.
PBR workflow tells artists that every color value should be between 10-240 to be physically accurate. Any color outside of that range is outside of reality. So if someone tells you “Pure red” should it be 240,10,10, or 255,0,0? And then I wonder if there is a curve in there?

If someone gives me fully saturated, bright, etc graphics they picked and designed in illustrator, when I bring it into photoshop (I do that Mostly so I can define it’s size in DPI as illustrator is still stupid and assumes anything you export is dimmensionless 72dpi) I’d love to just be able to add a saved “look” (even if it’s just one adjustment layer) to fix the colors to be within the realistic range. And I wonder if a curve is needed also, because of sRGB curves being abnormal, and previos workflows were gamma 2.2 vs the render engine which uses everything at 1.0, so you’d have to tell it your image was 2.2 or it’d be wrong.

I slapped my colorChecker passport on our scanner to get an even, single lightscourse image. That didn’t work out for me though, because X-rite lists the RGB colors for each of those, but Not the colorchecker Video passport. So I should be able to make a color swatch for each of those and overlay them on the image. I hope to be able to make as smooth layer adjustments as possible to those colors to try and get them to match the actual scan of the colorchecker, which would hopefully get me close to a color converter to “real-world” values? Does that sound about right? That’s about the best I could come up with. The only alternative would be for me to calibrate my printer, and print out the colors I need, then take a picture. Having to do that every time would be Very time consuming. If I actually have a pantone color, I could do the same, just take a picture under a controlled light, and get the color value.

BUT of course, I ran into a bigger issue, making me fee like a complete idiot. I noticed the ShareX isn’t giving me correct values. Then I noticed also that photoshop Also doesn’t give me correct values of screengrabs, and if I sample a color with photoshop outside of photoshop, it’s wrong. So I have some further research to do. My windows color management isn’t playing nice with reported values. Both values look correct. I pulled up this page: https://www.pantone.com/color-finder/2226-C and it shows 60,203,218,. but it samples at 142,196,216
The only way I can get the colors to actually copy/sample correctly is to set my windows color profile to sRGB rather than my monitor’s icc profile.

I did do one experiment with my color checker. It looked like DaVinci resolve might show you the actual RGB values after it corrects the colors. So I tried matching those color swatches to the original in photoshop. I ended up with a value curve that looks like this:
image
Where this is before:

image
image1135×673 48.3 KB

This is after the value curve applied:
image
image1140×681 49.6 KB

And then I added a selective color adjustment to get here:
image
image1138×689 50.7 KB

Does that clear up what I would be hoping to do a bit more? If you pick a color of a physical object on your screen, you’re basically making the same mistake lots of graphic designers make when they intend to print something out. Of course, once you add finish to the colors, you’re still making a huge mistake, because you can take a saturated red, then add a texture to it, and suddenly it looks pink in most light.

8

The question here is,what do you render?Using megascans?Or substance materials or which materials?.And what lighting?.Is it lookdev calibrated or artistic lighting and renderings?.Has it none contrast or any look applyed? ect.

Its very hard to give a general answer to your question how to make the renderings PBR pantone colors,if you have random renderings with differnt lightings ,materials and maybe grading or looks.

You could make a 2nd render with your colorcheckerchart in your renderscene and correct then the result with resolve or fusion or whatever .

9

Sure, for a colorchart. Yes. You could render a graphic of a color chart and fix the whole render in post. I agree. But if you Also had a graphic in the render, with the colorchart, and you fix the render with the colorchart your graphic color will still be wrong if you didn’t start off with it being the correct color.

What I’m asking is, how do you make sure the color is actually accurate? sRGB makes things really a pain in the butt because it’s not a normal gamma curve. Pantone is actually using sRGB values now, which is handy. And I know that because their swatches aren’t images, so when you inspect element on the webpage, you can see the values they used. And that’s great. If you get the RGB values from photoshop or illustrator, they are different.

And with PBR just Starting with the acceptable physical values of color being between 10-240, with suggestions for different types of objects being fairly different than that, I question what should actually be used for albedo. For instance, one suggestion for coal is around 52 as a value, where another is 20-something. And that’s an annoying one because the outside of coal is significantly lighter than the inside. And are they referring to charcoal, or coal? How many of us have actually seen coal in real life? Charcoal is super dark. Would 20 be the value of charcoal? My point is, would physical values of color need to be scaled to fit within that range? If so, is it a linear scale, a standard curve, or a special curve?

10

If you use Eg. Quixel megascans,they are photoscanned and autocalibrated materials.You can get photorealistic render results with them.There are some other PBR materials you can load in net.But Quixels scans are used for Movie blockbuster,Games,Product renderings etc.If you are rendering with such PBR materials and good lighting you should get a photorealistic result.

How you are postpro or grading your rendering then is up to you.

If you take a Foto with your camera its the same,today you can upload every jpg or whatever they accept to a printing service and you can get foto or poster whatever you want.They have automatic processes to convert the image to CMYK.

Afaik pantone is like a print ink.If you want to print your render ,you should ask your print service what is the best way for best result.

11

Pantone is just an example because most designers know what it is. It’s not just for print either. The idea is you’ve got a color on your computer that looks like the color of print or plastic. Others have samples for powder coats, metals, etc.

The point of asking about it though isn’t to print it the right color, or even look the right color once rendered, but Rather to send blender the physically correct color. The goal is just if someone gives me a spec and only gives me a color, I want to be able to send the right info to blender so that light, reflections, etc is calculated correctly.

So if I take a bright blue color and apply it to something, I don’t Necessarily want it to render out bright blue. If I have stylized lights and use a red, yellow, and purple light setup, Obviously that blue color is no longer going to be blue. But since the rest of the render will be tinted as well, it is still perceived as the correct material color (unless I went oddly extreme) because the viewer is looking at the entire image.

When I do this for a client in keyshot, often times the plastic is coming from china, we don’t have a color sample at all, so they use pantone as a target. I’m also using white lights for product renders, so generally I just put in the pantone color to start, then I take a swatch of the pantone color and overlay that over a good section of the render so that it visually matches. In those cases, they don’t actually care what color it is. They just want the final output to be matched to the color, and that’s basically what you are describing. Most clients I have do the same thing with photography even. They tweak the colors to match the target, even though it isn’t necessarily “correct.”
But if I have the entire model with physically correct materials, except I have no idea just for the logo, which may contain multiple colors, I’d love to be able to input those values with a much closer starting point. I’m just trying to work out a good PBR workflow, and this is one of the gaps. If you take a bunch of PBR materials, then you just slap a logo on it, the logo is Glowingly bright because the values are outside of the PBR range. The RGB values given are for appearing correct on screen, which is giving off light pointed at your face. And one big thing I haven’t really figured out yet, is how PBR colors are calculated. It seems like they are either meant to just be providing a starting point for artists to adjust by looking at the Output rather than input, OR to use actual physical scans. I’m just looking to see if there is a way to predict what those colors would be, if they were scanned, given a standard RGB value. In this case, I’m talking Pantone, but RAL and others would apply as well.

It’s on my list of things to do, to make a Substance Light scanning box (not sure what else to call it), and then I could just scan a few of them myself, compare with values, and probably figure out an approximate adjustment curve that I could apply that would work close enough. I was just hoping that already existed somewhere. Every time I think I have some time to build one, I get caught up in another task (not complaining about being busy at work compared to last year)

12

Ok in short i think this is the main question?What is the physically correct color?The answer is the albedo color in the PBR materials are physically correct.If you know how photoscanned materials works,with polarized lighting etc,you may not asking anymore about it,because its widedly use standard for PBR materials these days.

If you want a specific PBR color from a manufactur you have to ask or search for reference e.g. brand colors.

For cars its quite easy you can google a color code,same for a typicall brand color.

How you build your mat then is up to you.As example,if you want a plastic material with a color your have,you only need to add a clearcoat with a fitting roughness on top.

Read and understand these Guides and you know everything you need today about PBR

13

I mean… It’s not really physically correct unless it’s a full spectrum analysis. No such thing as physically correct color in RGB.

14

What you are talking about.We are using RGB albedos in allmost 99% cases?
I guessing what you are mean is a fully spectral rendering.Yes its more precise,but the RGB works today with most renderengine today or do i something missing?If these are Photoscanned and reflection polarized its working.

15

No, I mean if you want to render a physically correct color, you have to start with not only using a spectral renderer, but also feeding a full spectral analysis of each surface point to the shader. This is entirely impractical, but that’s what it takes to be physically correct.

I don’t really get the obsession with being physically correct. It doesn’t really make your images any better.

16

I think this brings more confusions then a solution to understand the simple PBR workflow we use today.

Or do you render PBR mats as Quixels megascans or substance or texturehaven and many more as spectral rendering?Even if you do,the source of spectral PBR materials are not availabe for the public if i am not wrong.

17

I most certainly do not. What I’m saying is, it’s pointless to sweat it too much as long as it looks good because actual correctness is out of reach anyway.

18

Sound to me as a artisticly building mats theory.Fair enough,but i make the statement that PBR rendering is PBR correct.Because its mostly photoscanned and calibrated with reflection polarization etc.It is limited to RGB in most cases,but it works…it simply works.

Something i like to add.
Even if you are rendering with the best spectral rendering engine today,the final result is for display at a RGB type LCD or OLED whatever display.

The only correct display i can think of is a laser cinema projector which could display every wavelength with its laser.And even this is maybe RGB filtered.

19

Well yes, that’s exactly what I’m looking for. But in my searching, I haven’t seen Pantone to list their PBR albedo color. They list RGB color, and they recently (Finally) changed them without saying why. I’m pretty certain it’s because they finally got around to adjusting them for gamma 2.2 instead of averaging to 2.0 since Mac used to use gamma 1.8.
Also in those creation guides you posted, that’s still not helpful for known color samples if you don’t have a physical sample (or a way to accurately scan the color value).
I haven’t scanned any yet, because I still have to build a scanner setup and also do a substance demo so I can see how they are converting the captures.
Even if you “scan” a color value, there are still different values based on if you are scanning at D50 D55 D65, and also what angle, etc.

Albedo values for non-metal (which should be nearly all pantone and RAL) are sRGB for screen. What I’m hoping to learn is, if you compare those values, is a straight conversion close?
So in the example of Pantone 2226 C, you get a linear albedo value of 44% (or 85% sRGB) if you do no conversion at all. But Pantone lists white as 255 and black as 0. Given that those values are out of the range, should I then clamp the values to be within the physical range as a starting point? The guide you sent says 30-50 and not to exceed 240. So if I clamp the output levels to 30-240 then the brightness goes down to 42% (82% sRGB). But the guide makes some of the things a bit confusing because it assumes your colors are scanned physical values. So if I clamp the value, would I do that linear, or with a regular gamma of 2.2-ish (since sRGB isn’t actually a gamma of 2.2 but rather 1.0-2.4 with a somewhat unique curve).
So I added an exposure layer and changed the value to .4545, then added the levels clamping to 30-24, then added another exposure adjustment with gamma 2.2 to bring it back from linear.
That Did make the color more dull, which is one of the issues you get when you try and use a pantone color as albedo. They are just Way too saturated, to the point that they nearly glow. But it’s also technically brighter.

You said you can easily google some values, especially for auto, but I’m not really finding anything so perhaps I just don’t know the search term to use.
I think I’ll just have to finish making the scanner, actually scan a few pantone samples in the dark, mid, and light range, compare the values that substance ends up with and the pantone’s given RGB value, and hopefully can just create some adjustment layers in photoshop that bring the values close enough that I can get a pretty good estimate of the actual physical values. I’m a Bit surprised that no one has done that already.

20

So the trouble with “if it looks good, it is good” is that only really applies if your goal is to make a good looking image. Rendering is often used for things like evaluation of a product design prior to moving to physical samples.
So a good point with Pantone (I really hate pantone by the way, for a similar reason to what you’re saying about Actually being physically correct) is that designers will often start with pantone color selections are a color sample target. But they’ll look at color breakouts for a product with something like Illustrator, on screen. I can put those to the parts in 3D, add some lights, and actually apply the finishes. One thing I often point out is that if you take a color like bright cherry red, but then you texture it a little bit, since it’s plastic which isn’t dielectric, the reflections are white. And what’s Red plus white? Yeah, you just turned your red plastic into Pink! If you scan that color with a full spectrophotometer, it’s red. But if you Look at it, with eyes, it’s pink.
So I want to show them their colors, With their finishes, so they can more accurately choose a starting point for the colors of their materials. We’re not often working with injection molded perfectly smooth plastic, so you definitely have to take texture into consideration, which you can never get in 2D software like illustrator.

And no, I understand that cycles, eevee, keyshot, vray, etc will all end up with different results. But if you start with all of your colors using the same PBR scale, then at least you should get close to a understanding the contrast you’re getting between the charcoal like black material and the bright blue graphic you’re trying to spec on top of it.
So my goal isn’t to make it look Good. My goal is to make it look pretty close to correct. I can worry about making it look Good later when the product is finished and released.

And even if you’re talking about calibrated displays, the point there is still moot because you should never Really look at physical colors of real things, and try and compare them to light on screen. It’s just never the same effect. Even if you had a theretical laser cinema projector that could display every wavelength, it wouldn’t matter because the material it is bouncing off would still be filtered by it’s material. TRUE physical correctness gets Really complicated because different wavelengths of light are reflected at different values depending on the surface material, diffusion, Angle, etc. So sure, I get that it’s not truly physically accurate. But it’s pretty close to Relatively physically accurate compared to itself, and that’s what I’m after.

The issue for me right now is I have RGB colors specified, and I have scanned materials. Those values are on different scales. And I’m having issues similar to what you would have had before the linear workflow existed. We just didn’t understand why, if you tried to use “accurate” colors, the render looked either dark, or Way too much contrast. Turns out that was because we were using RGB values, but everything inside the rendering engine was using linear values on those, so you had to convert all your RGB values to linear first (most everything does that for you now like in blender it’s sRGB for albedo, but linear for anything that represents a value).

So using the “if it looks good it is good” method for this type of work? Well, for one it’s inefficient. I have to manually guess on my colors every time. If you can actually come up with a process to convert it, you’ve just increased your speed. I’d say maybe there are artistic renders, photographic renders, and physical renders. I’ll end up on the artistic side for Some concept work, but most of the time, I’m asked to be somewhere between photographic and physical where they want it as close to a physical simulation as realistically possible.

1 Like