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
There are no special settings involved. I found out that strand based hair can look very good or somewhat meh - it depends a lot on the technique and how you model the hair strands. I only used the clump feature in the clumb tab and almost none of the settings in the tweak tab. I'll need to have s closer look at the parameters I chose for my hair model, but let me say this - it's not so much the clump and tweak settings that define how realistic the hair will look but rather how you model the hair strands. And due to the way how the editor works there is a lot if trial and error involved in this process because you don't have a lot control over the hair strands once you use the clump feature. And if you don't use clump, your hair tends to look bland. Of course, I really you model the strands first and then use clump, and don't go back to the style tab ... but I can't, I have to switch back and forth to define how my hair will look like. YMMV
I also used MMX hair shader. They add some extra feel of realism, but only the extra bit. They can't turn a somewhat non-realistic hair model into something very realistic. And also, if I used the standard dual lobe hair shader on my hair model it doesn't look too flat either. You can achieve very good results with the standard hair shader, but with the shader maps and the right light setting - I used an HDRI here - strand based hair can look very realistic and I wanted to demonstrate it. The same level of realism to match the level if realism you can find with bluejaunte's characters, and that's not easy to come by.
I'm so impressed by some of the work here. I can get results I'm very pleased with for black and white work maybe 60% of the time, as long as I'm using a studio set up, but I also go for high-contrasty looks that look very, very studio:
With color, I'm lucky to get anything approaching it less than 20% of the time. This one turned out as good as anything in my portfolio:
Most of the time, when I'm doing color, whether it's HDRI lighing or studio lighting, the skin requires very careful toning to keep from being flat, and then often looks overly detailed. Pore regularity often gives away the game.
I love it!! I was like, "is that one of mine"! Add some imperfections to the skin and you basically got it. Nice.
My latest attempt:
Thank you for your comment.
I will write about what I noticed while following your path.
My prediction: In Iray @ DS, only "scenes with one light" can guarantee "physically accurate rendering" (of course there is no IBL there, or the scene is almost completely surrounded by walls) And IBL must be almost completely minimized).
I thought that only when all scenes were lit with just one spotlight, IRay @ DS would allow “accurate” rendering that would be comparable to other renderers.
The hand looks more real than the face. On the face, it looks like the bump/displacement is too high. But it might just be the lighting. I've found—and it seems that most of the others have, too–that good results with HDRI lighting is significantly harder. Which is weird, since I think most of the Hollywood CGI people. Maybe, like @Junos suggests, Iray is optimized for a specific kind of lighting, which is closer to what @jeff_someone uses.
nice but how is it more accurate when there's just one light source? I though multiple light sources don't really interact with each other except for additively.
In response to davidtriune's question: When illuminated by a single light, the engine does not calculate anything about the interference between the light sources. Therefore, I think that the accuracy of the Iray engine is easy to ensure.
I'm just a "drunk oldguy who babbles at a italian bar", so I'm just saying what I'm thinking rather than claiming the right thing without any evidences.
As a personal impression, when the number of lights increases or when IBL is used, I think the output image will be "a picture like a super real illustration drawn with an airbrush".
I really like the look you have achieved here, what are your settings for this render if you dont mind me asking please?
Its ok Ive read further through the thread and see you have already posted them, thanks.
I think its safe to say a few of us here would buy a step by step video tutorial from you.
Out of interest ho wlong are these render taking and what computer setup are you using?
thanls.
Hey that looks good.
Thank you.
Basically, I just traced what was suggested by Jeff's writing, so I think that the same output will be obtained if we do the same.
Some new ones for kicks... finally a different model... went for a more full-figured, girl-next-door type... or tried...
Just for kicks and giggles, have you tried shrinking the entire scene, except the camera and lights, to get a more realistic light fall-off?
The light falloff seems to be inacurately more rappid (smaller too), than any real-world lighting. Thus, the constant "flash" like look of any lighting source. (Flashes have a sharp and limited falloff, giving the blown-out closeness with high darkness where light should be casting.) None of the actual numerical named settings act like the "formulated" names they derive from. Noticed when you actually type in exact values from the comparison reference sources which the formulas were originally created from. (Lux, lumens, falloff, etc.)
However, I found that when you shrink the whole scene, a little light goes a long way, and the lighting values become more realistic to the formulas. Yet, they are also still incorrect, when measured. (Measurements are done using pure white and black, non-reflective and full-reflective balls, planes and cubes, compared to actual photographs of the same objects, from standardized tests. You can find most of those test results online, if not a photo, than an actual "formula", that relates to specific lighting values, based on distance measurements.)
For example, your bedroom shots all look like they were shot using an i-phone built-in LED camera flash. However, the lighting values you have, I am sure, are more realisitc as values from a 5000 watt floodlight, which has an odd, unmatching fall-off of that LED light, instead of the 5000 watt floodlight, which should be blasting the whole scene in pure blinding light, with no apparent falloff, observable even if the scene was "light balanced". (It should appear almost laser-like, as if it were ray-traced, from a linear light source.)
I normally shrink everything down to 10% of the original size, since lights don't "scale up or down", retaining the falloff of the larger light, so it now correctly matches the falloff of the smaller items. (NOTE: The camera WILL scale, and throw-off your focal distance, as DOF does NOT scale with the camera, oddly. Your DOF will be off by 90%, past your camera, if you shrink the camera down with the scene.)
Interesting method, but are you sure this does anything?
Because I just tried this and the translucency seems to get really wonky. The skin on my figures at 10% glows red.
There was no appreciable difference in the fall off that I could detect.
I tried that too and got the same results: Figures glow red.
Falloff in iray is inverse square, as it is in reality. scaling the scene will make everything brighter if the luminosity isn't scaled, butt hat doesn't mean it's fixing some kind of issue.
if i scale down, reduce SSS by 1/2 and turn the brightness down by about 50% i get about the same result
informative post tho
Wow Jeff! Excellent as usual!
Awesome!
That is odd. Could be the auto-adjustment for light-balance that is messing with your results, as it often does. Remember, it's not real translucency, it's simulated translucency, based on "depth". If you scale your figure, I would imagine that the absolute values in light depth transmittance could be causing the issue. If it is set to transmit 2 units, and you scale it down to 10%, 2 is now 20 units of transmittance. (Eg, you turn the models into wax figures or semi-transparent latex rubber dolls, without bones.)
Was this on an IRAY material or 3DeLight base material with SSS, which is 3DeLight specific. IRAY uses "Scattering" and "Absorption", within the "Refraction" shader, which there is no direct relation to an SSS conversion from the old material format. The values for SSS would only actually apply if you rendered it with 3DeLight. Just as the scatter and absorption values will do nothing if you render with 3DeLight.
Also, what were you using for your light source? The default lights, from the top icon-bar, which are kind-of a hack for 3DeLight compatibility in an IRAY scene, or an actual real IRAY "emissive" light source? Or the scene-killing camera light. (Only actually useful to see items in the texture preview, as it has zero corelation to IRAY values when rendered, compared to what it displays on the screen.)
It was an Iray material, using emissive lights and an HDRI as the environment.
Would you mind posting an image rendered at 100% and one at 10%? I'm curious to see the difference.
Looking at the latest three images from @jeff_someone, it seems to me that the skirt is a give-away. The draping is all wrong - image 1 has the skirt appearing to fall through the seat geometry, poke through of the red chair surface in image 2 and the drape in image 3 is, well, just wrong. I suspect that either there is no dForce applied to the skirt or that the dForce settings are wrong. The skirt is just too stiff but the top looks ok other than I would always avoid dots as a texture pattern as the texture stretching is another give-away.
Yea the skirt is pretty badly done, the boots don't help, the laces especially on the 3rd image look really odd and they don't look right as they aren't 'tight' they don't pull the fabric at the holes.
The 3rd image also looks as if she is floating not sitting on the floor.
To be honest just my opinion but these are the least 'real' I have seen you post.
I did muck up the skirt --- though it was dForce'd -- just it doesn't work well when sitting on a chair... so it looks funky for sure!!
@jeff_someone You may try leaving some space between the chair and the legs, so the geometry doesn't intersect as it apperas to be in your first picture. You may also provide a chair animation where the chair slides in place to allow the draping to be simulated fine. Beware not to do it too fast or hiper-speed, it has to be a "realistic" movement.
Those renders look great @jeff_someone the skin looks very natural.
I've also had problems with dForce if the geometry (of the chair in your case) is not dense enough to provide a good collision surface. I usually find a way to strategically place a plane primitive with enough divisions to provide that collision and then make that invisible in the render.
Hello,
Could you consider doing a full tutorial or write-up?
I don't see a lot of the options in the render menu for what you're telling us to do.
I would really like to make something similarly nice. Could you be a bit more direct iwth how you're doing what you're doing?
Jeff, your renders look nice but I think that a big part of that are the textures. Which in my mind aren't real textures, just customized for your scene.
It's like taking a big resolution photo and converting it into a texture. A lot of the details are coming off from the texture. That is why they look like photos that are cropped into the images.
If you look closely there are a lot of small shadows on the face that would not belong there with the lighting (are just from the texture).
If with the same texture, you can do different angles and expressions and all look real, than you have, I think, photorealism in your scene and renders.
You can notice something similar in the facetransfer plugin textures. The character does look like what you want with the original textures but if you look at it without textures (or with normal ones) they look nothing like what you would expect because the geometry isn't there.