Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
That definitely made me laugh!
Rashad - you make some good points, the only thing that I would take up is you say that using gamma is a matter of taste in many cases. Of course it is, you can choose not to use it, but it is more than just a post effect, it affects the way fhat Carrara calculates light and surfaces, in order to produce a more realistic looking result. As such, it is far more significant than I thought it was, hence me starting this thread!
Joe - i'd happily look at your challenge but I am on holiday this week so can't render anything. I'd be interested to see your version too.
Rashad, thank you....
Finally, an intelligent discussion of the issues from someone with something intelligent to say, rather than just childish personal attacks by self-assumed experts. Thank you.
However, I do take issue with your point about what can be learned by attempting to replicate a photo. IMO, it is EXTREMELY useful if you want to learn the basics of light, shadow, color, and texture. EXTREMELY....
Case in point....
I posted a simple photo I took of a concrete bollard, shown below in the second image, and suggested some folks attempt to replicate it. Why? Because you can learn a ton from the exercise.
For example, look at the bollard, and its texture. It's concrete, which means what? It means it is porous, and appears "soft", and even glowing. It LOOKS like concrete. So if you want to learn about the texture, a study of that photo might lead you to wonder how you would reproduce it. And to reproduce it, how would you generate that concrete softness and "glow"?
Hmmm.........as you scroll thru the possible channels, what stands out as a way to do that? AHA !!!!! Subsurface scattering !!!!
Now, without examining the photo, 99.99999% of all hobbyists would slap a concrete texture image and call it good. Or throw a gray noise texture procedural on it. But only if you study the interplay of light off the surface does it occur to you to try SSS.
And if, like I suggested, you take the photo into PShop and do some color picking, you learn what the ambient colors are. The sun color, the reflected sky color, the shadow colors, etc.....and you see all the very subtle, intricate textures that are so important in "selling" the image to the viewer. Like the dirt and the thingamajig in the concrete expansion joint, as well as the silicone sealant around the base of the bollard. It gets your mind in the mode of noticing detail, and including it in your scenes and renders.
I took the opportunity to take advantage of my own challenge and generate a scene trying to duplicate the important parts of the photo, and came up with what I call an 80% solution....not anywhere near perfect, but the important lessons were learned regarding texture and SSS, outdoor lighting levels and colors, shadows, and even duplicating camera angles and tilts and focal length....
Anyway, the first image (strange how the forum software places the first image you load into the second position....) is my attempt to duplicate the major points in the photo. I used one distant/sun light and 3 spotlights, no other lights or ambient or GI or HDRI or gamma tweaks or anything else, and it rendered in 2:45. It needs a lot more tweaking to get it perfect, especially with the SSS glow since right now the texture is fairly suck-tacular. But I think it illustrates the point, and could pass for at least "reasonably" realistic. However, and I think our resident mason will agree, what my attempt is sorely lacking is that tactile sensation you feel when you run your hands over smooth concrete (or whatever that material is)...that soft, almost chalky sensation. Mine is far too glossy.
Anyway, my problem with ambient light, and not trying to replicate photos, is that it allows those without a lot of background to convince themselves that what they're producing is awesome, when in fact it is, well, less than awesome. But they never know, because they never consider what it "should" look like, and what it really takes to "sell" an image to a viewer. If they have no "standard" or guideline to follow, then anything they do is awesome. No rules, no guidelines, nothing to compare to. Which is fun and easy, but doesn't make for excellent images.
And the other thing with Ambient settings....you can get a much better effect using other methods than an Ambient light if you just know what you're doing. And it renders fast, and looks like light and shadow is supposed to look. But if you don't know the alternative methods, then Ambient seems to be the only option. Which it isn't.
Of course he is.
You're Evil!
This is interesting to consider. An unbiased engine should be exploring all possible light pathways such that over time the render gets closer to perfect. A gamma setting should determine the way the "perfect" values are displayed on the screen, it shouldn't alter the values themselves. Since Carrara is biased however, maybe the Gamma does something special?
I'd be interested to see two adjacent Lux renders for comparison. One with Gamma Correction turned on and one with it turned off, to see how gamma affects the way the unbiased engine calculates. Because in theory, both renders would be unbiased and both would be accurate even if one turned out to be much darker than the other. Because based on your description, Carrara likely treats Gamma almost like an undercover IDL intensity setting, potentially boosting the overall power of the indirect light but only when colors approach whiteness. There are a few examples in this thread of where applying gamma made an image darker, which is amazing because I've never seen it have that impact in any other software I've worked with, it always lightened everything in a way not so different than...ambience. Oh Carrara, full of surprises. But maybe Carrara gamma shifts lower range pixels slightly darker? Fascinating. This is when it would be nice if the coders would just pop in, tell us what's up, so we can all move on and get to rendering. Instead, we mere users will pontificate.
The results are compelling. Gamma correction seems like an essential step in Carrara and for the life of me I cannot figure out why. Good Stuff Phil!
Joe,
Nice job with this replica study. There are a couple of observations that pop into my head immediately.
1. First, is our natural tendency as artists to exaggerate natural effects. Most often I observe this with shadow softness. The real sun is just a dot in the sky and though it gives soft shadows most artists overdo it so they can "see" it as their mind expects it. Comparisons with real photos highlight this tendency well. Nice example.
2. Sunlight. Oh Sunlight. Most people underestimate just how bright the sun really is, so I think this example does a good job of matching this up. Nice work again.
3. Skylight is always going to benefit from some blue tint, specifically rgb 147 out the 0-255 range. I'd say the intensity of the skylight has been very well matched, just need the blueness.
4. The surfaces, all of them, look nice to me. I cannot determine why the planks are pink in the test example, but it does make it a little harder to determine how well the SSS is behaving and how it might be tweaked to get closer to reality or not. I'm also convinced that SSS wasn't really necessary, but to be certain I'd need to see two comparisons, one with and one without. These types of studies require lots of samples before any real patterns begin to emerge.
The issue with replicas, is that they are only effective in the most simplified situations like with this spire. While applying SSS to all surfaces in the current scene is feasible in this simplified situation, a more geometrically complex scenario will not be feasible to render with these settings. One issue is extensibility. How far can we extend certain approaches before they become unreasonable? This is of course to be determined on a case by case basis surely. But the idea of extensibility is why global lighting approaches are necessary so often.
I am curious about your gamma settings? Also, I figure that you have the ability to re-create this scene with gamma turned on or off, so long as the decision is made at the start of the project so that you can scale your light intensities accordingly?
What I like about your approach generally Joe is that you are less interested in formulas of 3, 4, or 80 point light schemes and more interested in finding the lighting that fits this current scenario is a naturalistic way. Pre-determined schemes often rob users of the opportunity to think critically about the current situation.
Hi Rashad and thanks for your comments.
I'm not the programmer so I can't be certain but I can tell you roughly what I think is happening under the hood, so to speak.
At the core of the program, all calculations are performed in a mathematical floating point space. With no gamma applied, all colors and textures are converted to this as they are, the calculations performed and then the image is converted back to the displayed rgb space again with no correction. I have seen this described by others who understand this better than I do as 3d programs being set up by default to calculate and display things "wrong". And as 3d artists, we have got very good at compensating for this by adding extra lights, adjusting textures, setting shadows to unrealistic values, and all the other tricks that people use to make their images look "right" again.
With gamma correction turned on, the core calculations are just the same. What I used to think was happening was that a gamma correction was being applied in the conversion from floating point to rgb, and I still think that this is the case - but it is not the ONLY thing that is happening.
When you set a color or use a photographic texture in a shader, you do this so that it looks right to you. But you are viewing this on a screen that has a gamma factor, that is, you seeing it AFTER a gamma has been applied. The additional step that I think is taking place is that an inverse gamma is being applied to the colours and textures before all the floating point lighting calculations are taking place. This converts things to their "true" values so that the lighting calculations are more accurate, and then applies a gamma factor to display the final image on screen.
This is the essence of linear workflow and if you are intersted I would recommend further reading, either the reference I included in my original post (which in turn has some references off it), or simply type "Linear Workflow" into Google.
I have rendered images in Lux and Carrara both with and without gamma correction with as close to identical lighting as I can. Carrara gets very close to the Lux images in both cases (although I am sure that Lux will deal with complex scenes and materials more accurately than Carrara). It seems the main difference is the defaults, with Carrara having gamma turned off and Luxrender having a default gamma of 2.2.
This is why I am so excited by this as I can now produce images that rival those I can produce in Luxrender, but far faster and including features such as Carrara hair which can't be exported to Luxrender.
Phil,
I think I understand a little better now. There is a two step process and gamma correction plays a significant role in each of these stages. It makes me question the image export options and how well dynamic ranges are retained. a 24 bit image should still store a great deal of information. An image exported as an hdri should remain quite rich. One could easily play around with the hdri to achieve results similar to Gamma Correction, or at least I'm assuming this to be true but have never tested it. I hope to play around with this all a bit more myself soon. Thanks for talking with me, Phil!
Wow, and coming from you that's...well....
Okay, anyway, I guess behind all of the attacks and stuff the bottom line is that all you guys are willing to contribute is more attacks...and some resumes....
I'm kinda surprised that the challenge suggestion of taking photos of yourself dancing naked and rendering clothes on yourself didn't gain more traction....How in the world is a request to see some of your renders, for a change, an attack? Oh... I... I see. I'm sorry.
Of course, everyone is encouraged to experiment and make your own determinations on what is necessary, hence the reason for my challenge. It doesn't really help if I sit here and do all the work, does it? :)
In any case, regarding SSS, take a look at the bottom photo in my previous post, and study it. It is clear that the surface appears to glow, correct? Some might find it subtle, and others might find it blatant, but it does appear to glow. And there is a reason for that. The material is porous, and has millions of microscopic pits and holes in it's surface. Which means light enters the surface, bounces around inside the surface, then bounces out (reflects) into your eyes. That is, by definition, subsurface scattering (SSS). Different surfaces might behave a bit differently with their SSS (skin, wax, etc.), but it's all the same principle.
I would encourage anyone interested to generate some with and without renders, and see if you can match that glow using any other method.
I've also taken the liberty of updating my render attempt in my previous post with an increased SSS Intensity and slightly decreased Diffuse Reflection to enhance the apparent "glow" and "softness" in the surface.
And regarding gamma.....
I think that it's clear that there are MANY MANY components to a realistic render. There are surface textures, and, for example, realizing that concrete "glows" and understanding how to reproduce that. There are lights, and REALLY understanding how to use the various lights and what they SHOULD look like. There are shadows, and colors, and details, and on and on... ALL of these things go into a "realistic", or even "reasonable" render. They are all visual components that allow the viewer to believe the image includes all the important components to that they can believe the premise.
Now, how many of us honestly have mastered all, or even a few of those areas? How many of us have even come close to a basic understanding? That's not an attack or belittling, it's asking for an HONEST assessment.
Until you are competent in most or all of those areas, relying on only one setting in one component of those areas to make your renders "realistic" is probably not the best approach.
IMO, if you want realistic, or even believable, you need to be able to accept that you are going to have to abide by some rules, and be brave enough to challenge yourself to really learn those rules. And you are also going to have to be brave enough to challenge yourself to do things like duplicate photos, even though it's difficult and will clearly show the world your shortcomings.
Again, gamma is a very small component of making sure your system and your renders are calibrated to your hardware, but in no way is it THE WAY the make realistic renders.
Okay - no. Not really. Subsurface scattering is light scattering within and through translucent materials. Pores in concrete does not make the concrete translucent, in the slightest.
But Carrara does come with a great SSS example in the browser if you have installed the native content. It's a candle.
Many people have remarked at how Carrara's SSS doesn't really work - well it does. But as to be expected, you'll need to put the light source pretty close - like the candle example - to see it really show off.
Gamma correction is one thing. Linear workflow is another. Follow what Phil's trying to say once. It's like GC @ 2.2 flips a linear workflow switch in the render engine.
Finally, concrete does not really appear to glow. The example shown is a whitening effect of the portland concrete being drawn to the smooth surfaces of the mold by water. "Glazing" is a term often used when you over-work your wet cement causing that white pigmentation to occur. This would be more the compared appearance of plaster, not concrete. Still, the glowing is actually the reflection of all colors fairly uniformly. No light, no effect. Glowing, on the other hand, will still show with the absence of an additional light source.
Okay, Mr. Beck. As usual, many here like to act like experts and make proclamations based upon nothing more than their word and lack of knowledge. Instead of trying to find ways to discredit me, as your buddy says, WOW US with your results in this challenge and show us how to get the texture without SSS. And maybe show us some expert information to prove your point.
Don't worry too much about mastering anything, just yet - unless that was your goal from the start. The OP of this thread was really trying to illustrate how to initiate a linear workflow in Carrara's render engine using a specific formula. Wanna have some fun? Go back to the first post. Read it and try his example. It really is fun. No mater how you tweak the HDRI, the linear workflow will have a different appearance that if you don't use it.
Yes, and that's what I said....."flipping a linear workflow switch" means that it converts non-linear textures to linear ones, which is exactly what I said, making sure all of the textures and hardware in your system are calibrated with a linear gamma.
Geez, dude, stop trying to discredit me and learn something.
Okay, Mr. Beck. As usual, many here like to act like experts and make proclamations based upon nothing more than their word and lack of knowledge. Instead of trying to find ways to discredit me, as your buddy says, WOW US with your results in this challenge and show us how to get the texture without SSS. And maybe show us some expert information to prove your point.
First of all Mr. Mamma2000, I'm not trying to discredit you. Just correct what you appear to be trying to teach people. Subsurface scattering, by your definition would be the blackness that one sees an a cave entrance on a bright, sunny day. And that simply is not true.
See for yourself.
Yes, and that's what I said....."flipping a linear workflow switch" means that it converts non-linear textures to linear ones, which is exactly what I said, making sure all of the textures and hardware in your system are calibrated with a linear gamma.
Geez, dude, stop trying to discredit me and learn something. Wow, get over yourself! I was commenting to Rashad!
thanks guys - now PM each other please .
Punch Maim?
Really?? So how would you convert all of that you described into Carrara texture settings? Can you do a render for us to show us how you'd match the photo of the concrete bollard using what you just said?
Absolutely! Are you kidding? I wouldn't use glow or SSS.
edited to conform to TOS
Finally Ignoring.
Sorry for the derail -
Of course, everyone is encouraged to experiment and make your own determinations on what is necessary, hence the reason for my challenge. It doesn't really help if I sit here and do all the work, does it? :)
In any case, regarding SSS, take a look at the bottom photo in my previous post, and study it. It is clear that the surface appears to glow, correct? Some might find it subtle, and others might find it blatant, but it does appear to glow. And there is a reason for that. The material is porous, and has millions of microscopic pits and holes in it's surface. Which means light enters the surface, bounces around inside the surface, then bounces out (reflects) into your eyes. That is, by definition, subsurface scattering (SSS). Different surfaces might behave a bit differently with their SSS (skin, wax, etc.), but it's all the same principle.
I would encourage anyone interested to generate some with and without renders, and see if you can match that glow using any other method.
I've also taken the liberty of updating my render attempt in my previous post with an increased SSS Intensity and slightly decreased Diffuse Reflection to enhance the apparent "glow" and "softness" in the surface.
We don't disagree on this point. I fully acknowledge that most surfaces will benefit from some degree of SSS, or dare I say it...Ambience. In theory pretty much any surface that has color has some degree of SSS going on. But just how "sub surface" are we talking? A surface that allows light to penetrate only a few atoms beneath it's surface is hardly a surface that requires SSS. Diffusion in graphics is supposed to be the standard representation of that very minimal SSS, and it also represents matte surface disruption like those pores of the concrete. That's why I don't think the SSS was essential, even though it may have been helpful. You are employing SSS to tweak the diffusion. But one could fake that SSS result with an Ambience Map, making yucky Ambience the hero again. One of the many reasons we cannot simply write Ambience off completely. Because effects are subtle, we don't always have to go crazy over them. There might be other more obvious details that are more distracting to contend with, such as the odd platform color and the missing blue tint from the skylight.
Ambience, for all its flaws, can be a quick and acceptable solution to that subtle glow Joe has rightly observed. One should almost always use a map of some sort to give the ambience a specificity of range so that it doesn't give itself away too much as ambience. It really is the uniformity of un-mapped ambience that makes it suck.
Let's first consider that this is probably a painted wood or stone of some sort, a highly opaque surface that doesn't really have much softness in terms of texture to begin with. It does however have a great deal of surface roughness, much more than the average surface. Using SSS to reproduce the bounces created by the extreme surface roughness is quite an interesting tactic. I'd even say it is successful in this test. I'm still very much thrown off by the pink coloration of the platform, however. Is there any particular reason for the apparent color shift? Is this a legacy effect of the SSS?
It's all fine, the SSS example holds in my view. I think your test is good. I would have approached it differently, but there is likely more than one way to skin a cat in Carrara even for something as subtle as this.
Further dissection. It is possible that part of the glow you are seeing is an ever so slight lack of focus brought on by a hand held camera in a world affected by wind and warmth. There will certainly be some real world d.o.f. involved. Perhaps some of the softness is partly due to a slight bloom effect within the lens? Either way, if SSS helps us fake it then all the better. But SSS doesn't get us fully there yet.
hi Rashad et al,
Sorry for butting in,
The human eye can only adapt to a certain range of tonal variation.
If I recall correctly, we have a large hole in the center of vision where the optic nerve enters and connects to the retina. We do not perceive the hole because the brain fills in the image, interpolating it from surrounding information.
yes :) in addition to the blind spot
we also have a network of blind areas caused by the retinal blood vessels.
having two eyes at slightly different positions allows us to fill in the blanks.
interestingly some people can see the shadows that the bv throw if subject to unusual oblique illumination
like opthalmoscopy
So... Gamma correction 2.2, then? ;)
I have no clue how or why we got diverted into a fairly irrelevant discussion of optic nerves and retinas, but last I recall we were waiting for Mr. Beck to tell us how to convert his explanation of Portland cement into Carrara settings, as well as provide us with a render showing the results.
Rashad, regarding your questions about the pink-ness or other colorations in the photo I provided....I'd suggest you bring the photo into PShop or equivalent and do some color picking to see what the actual colors are. I think you'll find that, within a point or two either way, the R, G, and B values for the entire image are pretty much equal, indicating most of the "coloration" is in fact just levels of gray. Yes, your mind and your monitor and your calibration of your system can affect how you perceive those colors, but the color picker never lies.... :)
Now the colorations in my attempt are way off, especially the bollard which is too red/yellow....
I'm also hoping that someone else will enter into the challenge and produce some results, but I realize that there is generally a lot of hesitancy when people have to match a photo. It becomes "money where your mouth is" time.... :)
And again, as far as gamma goes, believe what you want to believe. But until you really understand what "realism" is and how to produce it, and what gamma really is, all the talk of gamma is, well, just talk.
Now, as a public service, I decided to re-render my scene with a gamma setting of 2.2.....
And the resulting render is shown below, with the original photo for comparison. Now, I don't know about you, but it does NOT look even close to the photo. And that's because, with this particular system and scene configuration, what that does is apply a gamma correction to the surface texture which ISN'T NEEDED, because the original textures already have a linear gamma. But for those in the "I love 2.2 gamma" fan club, I'm sure someone will come up with a reason why it's awesome....
Are you serious? Personal attacks questioning my health?
Do you have NOTHING of value to add to the discussion, so you have to stoop to the very bottom just because you don't like me?
How about this: put your self-assumed brilliance to work and generate something of use so that we can learn from, not just childish tit-for-tat. Since you're supposedly an accomplished artist, show us some evidence of your skill and talent, and actually engage in this challenge rather than sitting on the sideline like the others and taking pot shots.
And you guys wonder why I criticize you......good lord, you should be ashamed.