[JCAddy]

Get a new computer

Looks good!

[superbad]

It's a 4 core Mac Pro!

[dilbert]

Wood with displacement grain instead of bump. This renders incredibly slowly.

Looks great, but you do bring up a good question. Just how badly does the displacement mapping really effect render times? It would be nice if someone could do a simple side-by-side render test of a simple object, first with a greyscale bump map, and then re-rendered using the same greyscale image as a displacement map. I haven't upgraded to V1.6 yet, so I can't do it, but some hard numbers would be nice to see so the user can evaluate what scenes are worth using displacement mapping on rather then bump maps. Something along the lines of "Diaplacement mapping causes a ...% increase in render time" in comparison to the same image used in a bump map.

[superbad]

See my edited post. It's about 4 times slower. I wouldn't consider this a good candidate for displacement at this point.

I want to try it again on a piece of furniture at a normal camera distance though- the reason I don't use much bump is because small scale bump really jumps out and creates a bunch of nasty, unrealistic highlights everywhere. Maybe displacement won't do that.

[JCAddy]

With any render engine comes a slow down when using displacement, why the shock here? There are even example renders in the manual that have render times stamped on them.


Also, I realize these are just tests that you're running but why would you displace a wood grain that small anyhow?

[dilbert]

With any render engine comes a slow down when using displacement, why the shock here? There are even example renders in the manual that have render times stamped on them.

I fully expect a render time increase, but it's the size of the increase that's important for my paying work (thanks for your example superbad). I'd love to use the realism of displacement maps, but it's a time vs. money equation. If the increase was say 100% I might consider it for a lot of my work as my dual quad-core burns through Maxwell scenes and it would still be worth my while financially. However, if it was a 400% increase in render times (as Superbad suggests), that turns a 2 hour render into an 8 hour render, which just isn't cost effective for my business. The displacement results are just outstanding from what I've seen so far, and I'd love to incorporate it into the pipeline, I'd just like to see some comparison numbers. If someone has the time to make a "Sample scene" with one using just bump maps, and the same scene using just displacement maps, and then compare the two at the same sample level, we could see if the quality increase justifies the extra time. Just a suggestion, and it would be interesting to evaluate.

[superbad]

Also, I realize these are just tests that you're running but why would you displace a wood grain that small anyhow?

I don't know, why not? Just wanted to see what it would do. This is my first time using a rendering software that even had displacement, so some of this stuff might not be as obvious to me as it is to you. Actually, I was (and still am, haven't tried it yet) hoping it would look better at longer distances. Bumped wood grain never looks right unless you're up close to it.

I'm not exactly shocked at the increase in render time, more like mildly surprised. And like dilbert points out, in this case it's not worth the extra time (to me anyway). There are plenty of instances (e.g. carvings), where I have long desperately needed displacement, and the time increase there will be well worth it.

[Mihai]

You should not compare bumpmap vs displacement, but rather real geometry vs displacement. If the look of a bumpmap and displacement is comparable, there is no question which would be most efficient.

It is also very difficult to give you a 'standard' estimation of render time. The height of displacement isn't the only thing that matters, but also the precision, and how much of a displaced surface is visible in the image space (fills up the whole image, vs a smaller area of the image). In this example the height was small but it looks like precision was very high. You could probably lower it quite a bit and get faster render times.

[dilbert]

It is also very difficult to give you a 'standard' estimation of render time.

Agreed, but I think you could do a side-by-side test showing the exact same scene, taken to the same Sample Level, comparing the quality of displacement vs. bump. For example, I've seen some outstanding grass, brick, and tile displacement materials already, so surely these would be great for a side-by-side test to show the difference in quality vs. the increase in render time. I'd do it myself, but I haven't upgraded to 1.6 yet

[Mihai]

But how would you do grass, bricks, tiles with bump? You can get a pretty good effect with Maxwells bump, but displacement is usually more suited for these things so the comparison doesn't seem valid to me.

[dilbert]

I don't know if I agree with that. Looking through the Maxwell gallery, I can see some outstanding architectural images that use Maxwell bump maps for brick and tile.

As an example, this outstanding material by Maximus would give a good comparison:

http://mxmgallery.maxwellrender.com/new ... =0&id=3203

It's not often that you would want an extreme close-up of a roof tile, so it would be nice to see a side-by-side of a material like this applied to a roof, one with a displacement map, and one with a bump map.

[Renato Lemus]

my test.... just a water bottle

http://pics.livejournal.com/macray/pic/000fsg19

Great example!

[bjorn.syse]

To get Bubble wrap right, wouldn't that require air inside the bubbles?

[Hervé]

To get Bubble wrap right, wouldn't that require air inside the bubbles?

for a physically correct bubble wrap.. yep.. air inside.. maybe instances..?

[def4d]

i think it needs two surfaces, one flat, one bubbled, or am i wrong?