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
@j cade She looks great. I'd like more realistic clothes as well, your top looks good. Textures can make a difference I think.
Love the hair. Too me the fine hair should be thinner.
@j cade When it comes to realism, the second convinces me more than the first one, I think the slightly stylized effect helps, especially the somewhat more desatured look.
Just thinking here... what would it look like if you removed the reddish undertone?
i really like these parts of the hair, the other side is okay but i think the drape is defying gravity a bit too much. Overall it's still better than most hairs.
I don't think cherry picking on the thickness of the hair at the model's hair line will bring us closer to how to achieve photorealism.
When you look for realism, you won't ever get it looking 100%
In my own renders, I just try to go as realistic as I can-- for 3D that is. I know even games that budget millions of dollars don't get super realism, so I try to get on their level of realism. It had me thinking though this thread, as my own stuff is what I want to create. I don't try to make it super realistic. But as an artist, what makes anything look realistic is the texture, and finer details of a face. I have Face transfer, and although it cannot render the side of the face, it can capture some textures nicely. I just really wish that at some point, I can add in an entire head texture, which will 'try' to copy the same texture for the side of the head as well. For a close-up face this would look good what I did here, but I will still work on the mouth in post-processing a bit to get that shine off the lips in the one render (if I worked on this. Remember, none of these has post-processing on them in terms of Affinity Photo or Photoshop used). I then adjusted it, added sun-sky for a light source (which is my favourite by the way), and then added hair on the face- eyebrows to compliment the existing eyebrow texture captured in Face transfer, along with actual long hair (reminded me of this series I am watching on Amazon Prime, so long hair it was). Lightened the sun a bit, and find that this character can pass for a decent 3D character in a game. I never kid myself to find realism, as in human-realism, as that is just silly. I am happy to have good 3D realistic, and that is all I can ask for in life. A character like this here, is on the same level of darn good artists that undergo university degrees at high class art institutions. So for me, this is awesome that I can to some degree get that same realism.
@nonesuch00 and @Leonides02
It's a personal shader I'm just using for my own renders. It won't go for sale :(
This is rendered in Blender. This was to just see if getting a default DaZ character to look as realistic as possible w/o additional morphs or changes to the texture.
@davidtriune In my experience the chromatic mode is buggy. Below an example where I go from 0.001 to 0 in the red channel of the sss color. You can see that zero switches something and the face tint goes from pink to green. I used the spectral rendering but it's the same without. This happens in many situations may be @RayDAnt that's the iray expert here could glimpse some light.
Staying away from 0% and 100% values for both the transmitted and sss colors seems to avoid these glitches.
Judging by the comments on this render, it would seem that @mal3Imagery nailed it.
no big deal. It's good to see blender is doing so good with realistic looking skin.
Thanks for sharing your settings @j_cade !!!. This is a new attempt, details on my gallery. Rendered in Iray Canvas, using an HDRI from HDRI Haven and Tone Mapped in Photoshop. No other post applied.
I transferred from the eyelashes (i assume thats what you meant) and it didnt work the first time, but for some reason when I transfer twice it worked. Thanks a lot BlueJaunte.
I was going for glossy lipstick, but I think the bump map has a bit too much small detail rather than bigger forms so it ended up just looking a bit noisy, If I had actually spent time on the skin I probably would've edited the bump map (though sidenote the tex itself is Gi 7 which is probably one of my fav textures, its bump map is generally pretty excellent) this is an old render of Gia in blender I think right when they added some skin feature that I was testing
And I dont know what witchcraft it is but the bump map seems like its more detailed than 4k should be
Yeah I deeply wish there were a way to vary hair thickness within a hair object (there isnt a way to do this in blender either and it makes me sad) It would also be great for memory conservation as I could make some of the hair in the underlayers thicker and need less of it and no one would notice
Yeah my continuous strugle on this was trying to find the balance of enough room for there to be layers of hair bending and falling in different directions and not so much volume that shes entering a beayty pageant in texas in the 80s I didnt go quite that far, but I probably ended up with just a bit too much poof
overall I'm just excited that there werent a buch of comments about how unnatural the shine on the hair is as I was unsure on that and tweaking the shader is a pita as the hair is memory intensive and cleans up real slow
super strange, but good to know. thank you.
I'm happy with gimp but thank you. I just wish gimp wouldn't be so darn slow lol
Okay Affinity photo looks way faster... just bought it. Cheers!
The other thing to do is have their base shape be just dots at the eyeline and then their actual shape be a morph (this is to some extent what the daz eyelashes do, thats why when you zero them they get small). So, since the initial shape is just dots all the verts for any individual lash all get painted the same weight and get morphs projected on them the same since all their vertices are in the same location
Mind you this method is dependent on whatever method you're using to create your eyelashes being able to scale down to the root. I use blender's particle instances, which has this option right there, curves in blender should also be able to do the same
I don't actually know if it works better or worse than using the lashes mind you
Also...
Congratulations Jeff! You know you're good when people start creating conspiracy theories about you
Lots of very good renders here! I just stumbled across @mal3Imaginery 's gallery render and boy oh boy would I like to get my hands on their uniforms! I have been waiting a while now for new military uniforms etc. to appear in the store so I could update this older render of mine. The Military Dress Uniform in the store is a bit outdated.
I think the surroundings, clothing etc. and extremely natural posing add greatly to the photorealiness of a render. Or distracts the observers eye from the obvious telltale features.
-P-
Thread has had its ravelled ends trimmed. Please keep the discussion civil and respectful.
My latest with brighter lighting and higher quality resolution (i.e. less 'old photo' style).
Mostly spitballing here, but it seems like it might be a colorspace translation issue between Daz's Iray Uber Shader and the Iray engine. Possibly caused by an overlooked/undocumented update to the Iray engine itself (would be far from the first time something like that has happened.) It could also just be an expected anomaly when certain edge-case settings are applied (such things aren't at all unusual where PBR engines are concerned.) I'll keep an eye out for anything that might explain it.
In the mean time I'd suggest bug reporting it.
That is...outstanding! Just...wow!
Now I know what some people have been talking about.....Spectral rendering and green lines at the UV seams.
yea, there was talk about corneas casting weird shadows if thin walled is off
you can turn it on which is fine,
but doesn't look good from the side because the refraction is turned off
In cycles it's all good
Fwiw here is the true explanatino for this phenomenon (taken from Nvidia's GPU Gems 3, chapter 14):
Dealing with Seams
Texture seams can create problems for texture-space diffusion, because connected regions on the mesh are disconnected in texture space and cannot easily diffuse light between them. Empty regions of texture space will blur onto the mesh along each seam edge, causing artifacts when the convolved irradiance textures are applied back onto the mesh during rendering.
An easy solution is to use a map or alpha value to detect when the irradiance textures are being accessed in a region near a seam (or even in empty space). When this is detected, the subsurface scattering is turned off and the diffuse reflectance is replaced with an equivalent local-light calculation. A lerp makes this transition smooth by blending between scattered diffuse light and nonscattered diffuse light (the alpha map is pre-blurred to get a smooth lerp value). Also, painting the stretch textures to be black along all seam edges causes the convolution kernels to not blur at all in those regions, reducing the distance from the seam edges at which the scattering needs to be turned off. This solution creates hard, dry-looking skin everywhere the object contains a seam, and may be noticeable, but it worked well for the NVIDIA "Froggy" and "Adrianne" demos. For the head models shown in this chapter the UV mappings fill the entire texture space, so none of these techniques are required. Objectionable seams are visible only under very specific lighting setups.
A more complicated solution involves duplicating certain polygons of the mesh for the irradiance texture computation. If there is sufficient space, all polygons that lie against seam-edges are duplicated and placed to fill up the empty texture space along seam edges. If light never scatters farther than the thickness of these extra polygons, then empty space will never diffuse onto accessed texels and light will properly scatter across seam edges. These extra polygons are rendered only in the irradiance texture pass and not in the final render pass. The duplicated polygons may need to overlap in a small number of places in the texture, but this should not create significant artifacts. We expect this technique would provide a good solution for many meshes, but have not tried it ourselves.
A more general solution is to find many overlapping parameterizations of the surface such that every point is a safe distance away from seam edges in at least one texture. The irradiance texture computation and convolution passes are duplicated for each parameterization (or many parameterizations are stored in the same texture) and a partition of unity—a set of weights, one for each parameterization—blends between the different sets of textures at every location on the surface. This "turns off" bad texture coordinates in locations where seam edges are near and "turns on" good ones. These weights should sum to 1.0 everywhere and vary smoothly over the surface. For each surface location, all weights that are nonzero should correspond to texture parameterizations that are a safe distance away from seam edges. Techniques for computing such parameterizations and their respective partitions of unity can be found in Piponi and Borshukov 2000. Given a partition of unity, the entire texture-space diffusion algorithm is run for each parameterization and the results are blended using the partition weights. This requires a considerable amount of extra shading and texture memory and is comparatively expensive, but probably still outperforms other subsurface-scattering implementations.
The takeaway here is that these discolorations at UV seems are an ever-present inherent limitation of the way texture space diffusion works in conjunction with certain lighting schemes that is highly exacerbated when spectral rendering is enabled (because of the increase in spectrum bandwidth.)
I have to say that I am impressed with your characters. Everyday people - just the kind I also try to fashion by tweaking morph dials on my favourite purchased characters. No glamour models here. So good.
@RayDAnt Thank you so much for looking into both the chromatic mode and the spectral seams issues. I believe your technical contribution here is extremely useful. As for the chromatic mode I'll report to daz and eventually let you know the answer.
I mean if I'm not deleted first by some moderator that lately seem to be very aggressive. So I have no idea what they may consider "inappropriate".
@jeff_someone Sorry it seems in this forum I can't give you my opinion unless I have nothing to criticize. That's not the case. I'm glad that you obviously had a chance to read my post before it was deleted and I'm also glad that you're working to improve your lights.
@Siciliano1969 Those seams also happen to me when by chance I change something in the material attibutes. For example if I copy and paste a color from 0.623 to 0.62 then the seams may get visible even if the difference is minimum. I mean the materials need to have exactly the same values. Then as @RayDAnt explains the spectral rendering may get things worse.
Excellent and thanks for the info RayDAnt!
Fantastic info Padone! I will check all the materials and make sure they are exactly the same values. My technique is to highlight all skin materials and type in each value such as translucency weights 0.80, etc all the way down to the trasmitted measurement distance and scattering measurement distance. In order to get the proper amount of SSS and light spread I use a value of 1.00 for the trasmitted measurement distance and 0.030 for the scattering measurement distance. I find SSS direction at 0.30 creates that skin softness I am looking for. Of course if I remove the spectral rendering then the seam lines go away.
@Siciliano1969 Be aware that daz studio will not show you the real values unless you enter them with rmb. The material slots always show rounded values.
Ah huh....good catch... DAZ Studio is trying to be what we call "furbo"