[simmsimaging]

Hey all - sorry to have dropped out of this one, but work has been insane the last week or so.

I have done a test with the same basic setup file and posted it here:

http://www.simmsimaging.com/CG/comparisons.png

The left 1/2 of each image are tiles made from geometry, and they have an identical material as the right 1/2, only without any bump/normal map applied. The right half is a flat plane with bump/normal mapping done.

I tested the same map used as bump and converted to normal map. I used them both at varying bump values to see what difference it made. The normal map is a bit of an extreme conversion but as you can see from the tests - Maxwell needed it to get close to the same effect the bump map gave me.

I also tested both the maps using Mental Ray. (values are somewhat different, but the bumping was much stronger in MRay so I had to dial the amounts down to get a closer comparison - especially with the normal map)

I think it shows that there is definitely something amiss with Maxwell's bump mapping. At first I thought the normal map was working better/not having the same problem, but clearly it was just a question of amounts of bumping.

The weirdness with the lighting I originally posted about is, I think, a problem with Maxwell and not a material issue per se - and I don't think it's just the way bump mapping works in a general sense. It does seem to be relative to how much bumping is applied. With the normal map it was fine up to about 80%, after that the oddness showed up. The maps both work fine in Mental Ray, and at lower values in Maxwell as well.

So - will NL weigh in on this one soon? I hope .....

[JTB]

So - will NL weigh in on this one soon? I hope .....

Well, they didn't for these 3 pages, I guess we'll hear from them anytime now... Maybe a support mail or a post in the bug list will help

[zoppo]

to be fair - nl says in the manual:
"bump is a very sensible parameter and standard values will go around 20."

and it has been mentioned that using 80 or higher will cause strange effects.

[KurtS]

to be fair - nl says in the manual:
"bump is a very sensible parameter and standard values will go around 20."

and it has been mentioned that using 80 or higher will cause strange effects.

This is probably a good rule, but in this material I'm using the bump value 99: http://mxmgallery.maxwellrender.com/sea ... =1&id=2165

[Tea_Bag]

Another problem is defining the correct grey scale colour or black and white tones to a bump map is there a rule on the shades of bump maps?

[simmsimaging]

to be fair - nl says in the manual:
"bump is a very sensible parameter and standard values will go around 20."

and it has been mentioned that using 80 or higher will cause strange effects.

I guess that would be fair enough if seemed consistent - and if the bump level of 20 actually seemed to do something like what I see in Mental Ray at the higher amounts. I can see that bump mapping is different in Maxwell than other renderers, but I'm not having much luck getting something similar without weirdness. If anyone can explain how in a manner that is predictable and repeatable I would appreciate the help!




b

[lebbeus]

I've done a search, here's some of the previous bump mapping threads (bump seems to have always been screwy)


http://www.maxwellrender.com/forum/view ... mp+mapping

wacky artifacts with bump maps (though these refer to pre-1.5 versions):
http://www.maxwellrender.com/forum/view ... mp+mapping
http://www.maxwellrender.com/forum/view ... mp+mapping
http://www.maxwellrender.com/forum/view ... mp+mapping

there have been others but I'm having trouble finding them

[simmsimaging]

Here's a couple more

http://www.maxwellrender.com/forum/view ... ap+problem

http://www.maxwellrender.com/forum/view ... sc&start=0

Seems there is some kind of oddness going on, and it seems that some people don't have the same problem, and it seems that some things work for some renders and not for others - which is why I was hoping someone could post a process/system they find works consistently.

It would be great if NL could at least aknowledge the thread and maybe save me/us a lot of time messing around with this problem.

b

[lebbeus]

some of the problems seem to be the need for a much more refined (heavy) mesh, while others are unexpected treatment of files (normal map problems)…maybe all will be fixed once MW has displacement??

[simmsimaging]

some of the problems seem to be the need for a much more refined (heavy) mesh, while others are unexpected treatment of files (normal map problems)…maybe all will be fixed once MW has displacement??

Hopefully - or will it suffer from the same problem? Hard to say without knowing where it's coming from.

I find it odd that NL hasn't really addressed this issue in any of the threads that I looked over. It just seems to drop off as each individual finds a workaround or gives up - unless I missed something?

b

[tom]

Hi Brett and dear friends,

The problems in your examples are all about the intensity. It's not possible to consider this as a bug because we wanted not to limit the strength. This means you're allowed to set more magnitude than possible for peak white. And when you do this, there appears flipped shading artifacts, inverse-like bumps etc. You may wonder if higher values causes problems why didn't we limit it to a proper range. It's simple, one might have a weak map (means subtle heights) and wants to have it stronger without going back to Photoshop and adding brightness to it. But this freedom practically caused problems now I see. Here are some examples with the provided maps:

.

Another question was the resolution of texture size and in opposite to suspects, it does not affect the look.



But the blurriness/crispiness of bump texture would surely make a different effect. Since bump and normal mapping is fake method of shading with perturbing surface normals, edges having soft transitions will help the thing look better raised or inset. This is quite normal. On the other hand that wouldn't be problem for displacement of course.

So, how should you set the strength for staying in valid ranges? Easy!
Check your bumpmap's histogram for peak greys. If you have a normalized map, you'd surely have white pixels and this is very common for best bumpmaps. Then, I would suggest you not passing value 50 for strength and 30-40 would be ideal. If your map is darker, that means it allows you boosting it and you can set 100 for the maps having maximum 127 grey in histogram. More than these values will end up with artifacts on your render.

Same rule applies to normal maps as well. But how would you know how strong your normal map is. That's practically pretty easy because normal maps having peak strength always have most intense red/green coloring on the blue base. You can again check this by looking at the histogram per channel. I would again suggest you the same values above but in opposite to common bumpmaps, most available normal maps are weaker than their maximum possible values, so most of them would tend to allow even 80 or 100 as strength. You should know your values before to start or you can quickly check with an easy scene like this.

[Bubbaloo]

Awesome, thanks! Did you mention something about displacement?

[simmsimaging]

Hey Tom - Thanks for the clarification. I'd almost given up hope on this thread

If I understand you correctly I am not sure I'm 100% happy with the end result though, which is unfortunate. I understand not wanting to push the bump mapping so far that you get artifacts, but without displacement I think we could use a bit stronger bump mapping than this. Clearly Mental Ray is capable of stronger bumps without this issue, but I understand that it operates very differently and realize that does not mean it can be done in Maxwell.

I'm guessing there's not much to be done about it - hopefully we can get around the need for stronger bump mapping when the displacement function is released.

Thanks again for taking the time to go over it.

b

[tom]

Awesome, thanks! Did you mention something about displacement?

Brett, it's exactly how you said but that's not impossible to improve this situation. We have some ideas but not for the nearest update I will let you know more with comparison images when it's ready...

[WillMartin]

Hi Tom

My map was as normalized as possible; much pure black and much pure white, but each area of bumping looks different from the other in the end due to the different way Maxwell looks at gradating in the map (compared to LightWave).

The rule I have for myself at the moment is (and I'm open to refining this if I'm off a bit ):
If my map is doing a lot of big leaping in value from the dark pixels right to the light pixels then I use very low bump value in MW (less than 20) to get an equivalent to LW's 90% bump value; if I have huge amounts of greyscale between my dark pixels and light then I can safely increase the bump value in MW to much, much higher to get something like LW's higher bump map values.
And the weightmap approach that lebbeus showed us previous page seems like a good solution for maps that, for whatever reason, have both smooth grading as well as big straight jumps from dark to light pixels.

The only continuing problem area for me where the above line of thinking is helping was the airbrush dot and ring (see my previous posts in this thread with images). In MW it was difficult to get those to looks like LW's interpretation regardless of the bump value setting in MW.