this one is a surface-study, sculpted in ZBrush. The main purpose was
to see how much of the sculpted detail is visible when rendered in TG.
I just did a simple coloring to get a basic stone color, no displacement
or bump, so that the details are still recognizable easily.
The model is rather midpoly and thus not good for close ups.
Also the terrain is left pretty basic and simple.
Made a lot of different views with different lighting, but don't worry,
I'll just post a few.
And one to show the back.
Wicked model. Looks pretty good. Maybe the textures could be higher res.
Compute Normal and a Texture Coorrds from XYZ after that displacement will make any texture look good on that. 8)
Great model. Indeed, some extra texturing in TG and it can be really realistic.
:) Thanks guys.
Zardoz, the one with Sean Connery? ;D Saw it decades ago, so perhaps unconsciously.
Maybe I should have been clearer about some things.
The model does not have normals, so compute normal and texcoords from xyz won't have
any effect, I'm afraid.
The coloring is just one Power Fractal(world space) multiplied with a faint AO map
and an also faint cavity map (combined by a merge shader). Both maps are 4k, which
should be high res enough here.
Of course I agree that more PFs and, from my point of view a painted color map would
improve it. Tested some stuff, but it made it harder to see the smaller sculpted details
and those are what I wanted to be clearly visible in this case.
The AO/Crevices maps look blurry to me. 4K isn't really large enough IMO for TG. Always noticeable issues in my cases. I use 8-16k.
If those are 4k maps there is probably stretching?
4k should be good enough for that size (as you said) if the maps are clear enough to begin with.
The overall texture is a bit soft, but that can be overcome by procedural of course. If you have blurry AO (which I don't really see as obvious here), you can take it through a surface shader and have the greys broken up a bit as well, makes it less blurry/soft. Lots of potential anyway.
Whoa... I had a completely different thing in mind when I read the title. Didn't expect it to be so literal. :) Very interesting.
To me the 4k textures don't look blurry and the only stretching
is on the invisible bottom of the model, 'cause I shrunk the bottom
UV island to save space.
The PF is stretched on the y axis though, so that might cause the impression.
Created an 8k AO map last night and set up a little comparison for us to look at.
It is just the AO map with no PF color, only the default color of the default shader.
Rendered with a cache file to have no lighting differences.
Have a look for yourselves.
No siginificant difference. As what I thought.
I wouldn't think 8K is high enough either (think 8k pbr materials, doesn't look great in TG).
The geometry of your model the AO is coming from is pretty simplistic as well.
It just doesn't seem like AO more than it is smudging of crevices. There is no definition regardless of angle/shape.
In fact it reminds me of a crevice map with blurring turned up like you can do in Materialize.
I do also have a rather large 4K ultrawide true color monitor. So I may be seeing things differently at larger scale.
Meanwhile I tried some texturing changes, as suggested by Dune.
The first shouldn't be as soft anymore.
A bit more color.
And some scale change compared to the first.
What do you think?
WOW!!!! The last one looks best to me, but the other ones are great too! I really like the model, and I really like the cracks a lot!!
It's looking very good. They all are, I think. But I would add just that bit of bump, some mix 1 of a few centimeters, see what that does, up close.
Fantastic work. This is really developing in the right direction.
Thanks to all of you. :)
Hannes, glad to read that you like the cracks, 'cause they were one of my main goals here.
Ulco, I won't try bump on this model. There is some small sculpted stuff on this one, just
not for close ups. Maybe I try another one later with HD in ZB.
Close ups show a too facetted surface, because the model doesn't have normals. And with
normals - tried that earlier - some of the details get lost.
Hope that explains a bit.
Weird that details get lost when you have normals. I would love to see the differences, with normals, with a bit of bump, but it's your experiment.
Ok, ok, I'll try to show and tell you more on one of the next days.
The moss/algae version looks the best imo. Also curious about the normal, seems strange. Unless it's a texture filtering thing in TG?
From this series I prefer the last one!
Well, now some more info for you Ulco:
First let me show you a ZB-screenshot that shows some details of the surface.
You can see them here a bit clearer than on TG renders.
Here are some comparison shots that show the differences between
no normals (or normals without using smooth normals) and normals
with smooth normals on (should be one of the main reasons for using
normals on imported models).
You can clearly see that everything has a rather blurry look with 'use smooth
normals' on. That way you loose some sharpness and jaggyness.To be noticed better
on close ups of course(just like the polygon-edges). Some details of the sculpt
get lost in softness.
That alone could have caused my decision.
There are at least 2 more effects that don't really improve the look of the model.
The cast shadows are almost the ones from the no normals or the not smoothed normals.
Related to the already discussed shadows on too low poly objects I assume.
The (ambient) Occlusion weight of TG's Gisd also uses the unsmoothed polys effect.
So that makes close ups not the appropriate choice here.
Now for the bump stuff.
First another close up and then the same with the cavity map i used.
The 2nd image shows that more or less almost the whole surface is covered.
Appling additional TG bumps would cover or destroy the already exsisting
Furthermore the TG bumps and dents can't be integrated in the cavity mask.
So, if I would like more or different bumps and stuff I'd apply those in
ZB, because then they could be integrated in the cavity mask.
You can apply perlin or voronoi and other noises in ZB.Wanted to do that
originally, but changed my plan while doing the sculpture.
So or so my fix goal for this sculpt was to just apply some PF-based coloring
in TG (and of course some specularity and reflections).
Hope this explains it better and in case you still have questions, feel free
to ask again.
Thanks very much for your lengthy explanation, Jochen! I see what you mean. The smooth normals indeed makes the texture too smooth. But what about using a less smooth calculation, like 44.9º or 89.9º, or indeed, the default 35º of Poseray? You may keep the larger areas less smooth, but get rid of the small poly 'artifacts' in the crevices.
I may have to review the way I do my clothes, after adding them to the figures. I usually recalculate using smooth normals, so that may be not the best way. Though bump and detail comes out fine, I will test some versions.
I guess from your description there are normals anyway, but just not the smooth ones? If I make a sculpt and don't recalculate in Poseray, the object looks horrible in preview in TG, no shadows, or any curvature can be seen, all just one flat 'thing'.
Usually ZB does not export normals, so when I say no normals it really means that there are
no normals at all. You can see that in the preview of TG, if you remember, the whole object
is dark, no details visible.
(In case you have normals on your model, but have unchecked 'use smooth normals' in TG
it should still be visible in the preview, I guess)
Since a few versions you can export 'smooth normals' from ZB, they added some modeling
features with creasing and stuff and therefor many users needed such a feature.
There is no smooth calculation feature in ZB, as far as I know, so you would have to apply
creasing to edges, which can be a really painful work on sculpts and would not be really
good unless you would change some geometry as well etc. .
Doing that in Poseray wouldn't help either, because you would need more than one angle
on a sculpt usually (had that problem even on some Wings3d models btw.,which is the
reason why I recommended creasing edges in Wings as a better method earlier).
Some of the poly-artefacts in the crevices you noticed are caused by the triangulation of
TG (same would happen in P-ray), because it's an automatic feature.
You only can get rid of that by triangulating some features manually, but that would be
another royal pain on high poly objects (even on rather mid-poly stuff like the sculpt of
this thread, it has ca.1.5mil polys).
Yes, a no-normal object in TG is quite useless in preview, though it often renders fine. The thing with Poseray is that after importing a previously smoothly calculated object, it seems as though the smooth normals are gone again. Even after a recalculate in PR and importing again later (for adding parts to it, eg.) there's often the necessity to recalculate again.
The poly-artifacts in the creases are not a problem if a bit of extra bump is applied, you'd hardly or not see it.
Strange PR effect you are talking about.
Maybe there is a bug or so in one of the newer versions.
Never experienced that, but I haven't used it often since some years now.
So unfortunately I can't help you with that, sorry.
No problem, I just recalculate. I've used quite a few versions, but vaguely remember I reverted back to an older because the newer had something that didn't work for me. But this is a good reminder to check on even newer updates, thanks.
Nice model. I would definitely go for at least 8K or higher, or multiple UVs. The moss version looks best IMHO, given the surroundings presented.
Ulco - Thought about it again and it might be logic when you are adding some
more geometry to your model, but in case you just import a model again without
having done some modifying in another software and saved it with that, it would
be really strange. If you saved it with another software inbetween it could be a
different obj writing that can't be read by PR.
Just a thought that came up.
Thanks for posting all of the explanations with the examples. I'm following your progress with some interest. :) I really like the way this is going and I'm using 3DCoat to try to achieve some similar things. I'm curious how you are setting up the cavity maps in the TG shader tree? I've never had much luck with those in TG. I do like the cracks. Keep up the fine work!
Quote from: zaxxon on July 23, 2020, 01:07:53 pmI'm curious how you are setting up the cavity maps in the TG shader tree? I've never had much luck with those in TG. I do like the cracks. Keep up the fine work!
Cavity map are a form of AO, so I believe in TG we'd want to multiply with a lower value you'd typically use for an AO map. Maybe add multiplied colour?
Thanks Jochen. I will have to check what really happens and may make some screenshots, but the vertices looked 'harder' after reimporting a saved obj, without changing the sculpt. Just adding stuff to it in PR.
If I make a crevices/AO mask in ZB, I always check how soft/grey it is in Photoshop (not completely black, but a nice gradient), and multiply colors in TG, or add an empty surface layer, mask it inversely and use a color adjust as child for manual gradation of the cracks/cavities.
Ulco - when you add some stuff in PR (more separate geometry parts) it would
be something new and would have to be recalculated in case you added it to
the object, because it would be a new part of the previously model, I guess.
zaxxon - Thanks Doug. And now for your question.
If it the cavity-map was the only map to be combined with the color-map (image)
or the color done by PFs I would of course multiply it as usual (with a multiply colour
node). In this case there is - as you most likely have noticed - an AO-map and an
additional cavity-map. So this time I combined them a bit differently.
The transform input shader is set to 'Use world space' so that the PFs are applied correctly
over the the UVed Model.
The AO- map and the cavity-map are combined with a merge shader in the default mode.
Which means the merge mode - merge colour is checked and remains set to mix(normal).
With the mix control you can balance the two maps, either have them equal or one stronger
then the other.
It is done like this, because multiplying both maps leads to a somewhat darker result and
that is not what I wanted.
Have a look.
With the merge shader
With a multiply colour
The same on the colored model.
And since I used rather grayish (no black) maps it was much closer to what I was after.
The whole shader tree - without the greenish stuff - looks like this btw..
Hope that this answers your question. Keep on more questions in case my answer didn't
Thank you for the detailed explanation. In some other apps the cavity map adds additional detail to the AO map for the deeper 'crevises', and I guess this is what you're doing here in a TG way. I'll give your set up a go. Thanks again!
You can use all sorts of additions to setups like this, like a bias or gain scalar to influence the gradients in AO/cavity maps, or change the RGB/grey values in the image map shader itself, or multiply colors/scalars from the image maps by constants for more or less impact, etc.
Never tried bias or gain scalar with image maps, might be worth a try, thanks. :)
Quote from: Dune on July 25, 2020, 02:22:44 ambias or gain scalar to influence the gradients in AO/cavity maps, or change the RGB/grey values in the image map shader itself
This is what I gotta do with most shipped maps as they do tend to come with gray tones from generating rather than baking from 3d. And in apps like Materialize, I can't actually seem to find a exposure/gamma type setting to whiten what it generates when I try to make a new one, so end up going into photoshop after that, and it's tedious.
Just in case someone is interested, you can see a few more views and ZB renders
on my blog now.
This whole concept is great experimentation and conclusive results! Thanks! :)
:) :) :)