Thanks for the feedback Rueben. I had to take the videos down because Apple (.Mac) said I was using too much bandwidth.
I’m pretty sure the flat was because I changed the levels of the displacement maps in Photoshop. I know I upped the constrast inorder to get stronger displacements but I probably should do that with incremental test of Bump/displ. setting directly in EI.
I will check and eye tho and see if I’m in error.
Also, I just did sword render in GUV and it’s gorgeous. However, If AUV is working, (and by your tests, they are) I need to get that under control as well.
AUV as in Adaptive. Not Auto. I’m on the same page now. This is my first AUV render that works. It looks good and another option with benefits of texture resolution based on polygon size. So it’s Zbrush thing that promises no stretching. I think it’s good if you use Zapplink for Photoshop editing of the texture, then touch up whatever projection smearing it makes in Zbrush interface. But at least, I have finally warmed up to it
I’ll do more testing with this as well but with my own modeling.
Its started happening with every new project i create, i’ve got no idea why its happening, i’m doing nothing different, don’t get it…
As mentioned, i increased the bump and displacement for a previous “working” project to a really high level as a stress test and it displaced perfectly, the project below is “rock solid” -
Sorry, but for us it’s absolute not clear WHAT (Communication Breakdown). Where is the prj? Original map? Settings? IMO it’s better to learn a single prj in all details instead of running thru numerous lizards. Up to now we cannot support the discussion cause we see no facts but assumptions only
Whether its discussed here or not is of no importance to me, what is important IMO is that these inconsistencies with displacement are sorted out, this is the very reason i gave up with these “tests” in the first place.
Alonzo stumbled upon this problem with the “cube” project as shown in his image.
If its not a problem with EIAS then someone say so, cause i am unconvinced that EIAS is not the problem.
Now AFAIK its up to EI technology and its sub-contractors, i’ve done enough and had enough.
Hey Rueben…Not exactly where to answer this because I just described the problem here in this forum.
Yes, it’s the first time I got banding. I believe it is as I just mentioned with low fidelity image in directly generating an 8 bit from ADE.
It’s banding, due to lack of gradiation in the grayscale image map. So, 128 levels of gray in 8 bit isn’t enough to describe the smoothness of image as a 16 bit or 32 would.
Then, I believe the problem maybe be compounded by lack of AA or Blur.
I can’t be sure of this, haven’t really been infront of the computer for a couple days. (not much
Hey Rueben Sorry been having a couple of days fun in the sun… to feel human again.
Sorry I couldn’t get back to this. This is why…and oddly enough, keep photoshop in the mix (process) would be still be a good thing. So if you keep the 16 bits as long as possible it’s good. but if you go 8 bit, with no blur or AA or room for EI to smooth out the lack of data in the image, I believe it will band.
In the old printing world, 8 bits were never enough to describe a some graduated tone. Some of the other setting need to be available if you use them…like Blur. Or some other compensation in the render to smooth the image.
At least that is what I posted last. “The need for 16 bit will eventually become aparent.”
When I get a moment I will instigate your ADE script.
Hey Rueben...Not exactly where to answer this because I just described the problem here in this forum.
Yes, it's the first time I got banding. I believe it is as I just mentioned with low fidelity image in directly generating an 8 bit from ADE.
It's banding, due to lack of gradiation in the grayscale image map. So, 128 levels of gray in 8 bit isn't enough to describe the smoothness of image as a 16 bit or 32 would.
Then, I believe the problem maybe be compounded by lack of AA or Blur.
I can't be sure of this, haven't really been infront of the computer for a couple days. (not much :)
Hey Rueben Sorry been having a couple of days fun in the sun... to feel human again.
Sorry I couldn't get back to this. This is why...and oddly enough, keep photoshop in the mix (process) would be still be a good thing. So if you keep the 16 bits as long as possible it's good. but if you go 8 bit, with no blur or AA or room for EI to smooth out the lack of data in the image, I believe it will band.
In the old printing world, 8 bits were never enough to describe a some graduated tone. Some of the other setting need to be available if you use them...like Blur. Or some other compensation in the render to smooth the image.
At least that is what I posted last. "The need for 16 bit will eventually become apparent." I know you also want super crispness but without micropolys...it's going to be only so sharp on detail. Without 16 bit or micropolys, efforts to sharpen the image map with big out flaws. I feel 8 bits with ZBrush is doable.
When I get a moment I will instigate your ADE script.
You did a lot of great work...don't give up.
Thx, clear now. Yes, with such textures (like in ZB Banding) the “moire” is what should be rendered. In black areas RGB values (0…255) are like a lot of zeros with seldom 1-2 here and there. That’s enough only to get a moire we see. Your maps are sorta “terrains” - micro and macro are in one, that’s not for 8-bits.
About positive/negative (2 maps)… we don’t understand what do you want to achieve. It would be clear if you’ve 2 original maps, but without it… we’ve no idea, please explain.
Hi Igors, forgive my “tone” earlier (had a hard day).
Ok it would seem that in areas with no or little displacement/bump info this banding happens, yes ? this is the only difference i can see between “ZB banding” and “lizard head” projects, lizard head is almost fully covered with a bump/disp.
Maybe this is not such a problem.
The provided maps were the original exported ones, as we know Zbrush uses a 50% gray mid-point, these maps were generated with the ADE displacement exporter.
The positive/negative maps provide a solution for “standard” renderers where by 100% black = no displacement, the positive map contains the outward displacement info, negative = inwards, this has worked very well for me with AUV tiles without any of the “lines” around the transition between the 2 maps (like you seem to get with GUV).
The calculation of perturbed surface normals (bump/displ) is same anywhere, it’s based on math from times of 17-18 century. The calculation is very simple (in principle) and (to avoid math/geometry specifics) can be formulated as: perturbed normal is a texture’s difference (gradient), not its absolute values. So, just see your textures: are there nice greyscale transitions/gradients? If yes, than a good bump/displ is guaranteed. Note that gradient’s smoothness is much, much more important than gradient’s amplitude/range. A map can be very dark or very bright overview - it’s not a big deal, the importance is only how many half-tones you have. If you see almost no transitions (as in ZB banding), so you’ve no chances to see rational bump/displ in these areas. It’s same like if you increase dramatically an amplitude of dark texture areas in PS: yes, you’ve got more white, but… very rough, no gradient, no data and any amplifying can’t help.
First off, there is an popular contradiction between “artist’s language” and “technical language”. Artists: “I’m not interested in bump maps”, “Don’t worry about surface normals” etc. For artist it’s fully obvious (and self-clear) that if a surface is pushed inside/outside, then we must see it visually (darkness in pits areas for example). But technically “vertices positions” and “surface normals” are handled individually, absolute not automatically (typically 1% of work for “positions” and 99% for “normals”). If surface normals remain unmodified, you can’t see pits/hills but only a strange “ice” (just specify a much less bump than displ to check). Any modification of geometry (displacement, deformation - anything) should be “confirmed” by surface normals modifying, otherwise result = absurd.
So, all these 0%, 50%. 100% (or any %) change nothing in surface normals and result’s “mimics” (very convenient artistic term :)) remains absolute same. You just expand/shrink a model, but nothing more. Of course, it’s affair of artist how many maps should be used. Just it’s more rational/practical to start from a single map, be sure it gives a good gradient, and then (if needed) switch to 2 (or more) maps.
Yes, Displacement is cool but there’s some drawback with settings that directly generate 8 bit. I still export 16 bit. I haven’t tested it enough tho, so maybe there’s something that works. But I still hold on to the 16 to 8 bit conversions as long as I can during the process.
If this is directly related to the ADE posting in PostForum. I would make a correction or addendum.
Very good Igors. This is why I got some flat areas when expanding gradiation to a broader gamut. Because though range doesn’t matter gamut does. So I learned better ways to expand the range without losing important parts of the gamut. This is why photoshop is still good in the process, tho’ a bit more time consuming. As an artist I would like to reduce that time. I still haven’t used curves in Zbrush directly yet.
Again I agree Igors, bump is very importand to a good displacement and the two are co-dependent in most cases. ADE is good because it can export a full active sets Bump, Disp, and Normal. I aslso think there is another type, called caveat. Depending on the project you may want to be very picky of each map individually and lots of time perfecting each or just need a fast export pipeline. At one point I did think using more maps could smooth the look of surface but I can’t substanciate that. However, tho you could smooth banding with a blur in PS, I would perfer to get the smoothest range then leave blur in EI render. Don’t want to get into all the funky insufficient tricks to fix a stair-stepping map.
Yes, start with one map with high depth, separete or correct gradient range, then split, then reduce to 8 bit. So making all edit before simplifying. I would rather have a small range then up the levels to a more dynamic range in PS. Then only cut the lower value because “artistically” darkers areas are less distinguishable.
One question, Igors. If I shift the nuetral area and move the 50% nuetral to 0, what happens to darker areas? It goes to negative? And I only have to export two maps with no separation? Interesting thought?
No, there is nothing more that can be done with that exporter to improve the results, it has a “smooth” setting which seems to apply some amount of blur, but i could not see any difference in the rendered output.
BTW, i’ve looked at a lot of the displacement tests on zbrushcentral.com and a lot of the other renderers produce similar banding effects when 8bit maps are used, obviously there is only one solution to this :rolleyes:
Alonzo, you are right !, PS conversion to 8 bits gives much better results !, but sorry i’m not going to re-do the ADE expoter codes, no i think Pixolator shoud be made aware of this, the whole point of the ADE is to improve workflow, i’ll make some comparison shots and post em to ZBC.
