appartment vray1.46.05 testscene - free download

[flipside]

I hope this is just the start of the testing, so for me, they can keep it for a while and post the final extremely well tweaked result

bye,

wouter

[vlado]

But actually while computing for example first bounce IR map, it uses the info of the already calculated second bounces lightmap?

Yes.

If you set second bounces to qmc, will this slow down both the IR pass and the actual rendering pass?

Just the irradiance pass; once the irradiance map is computed, it doesn't matter what is the secondary GI engine; for example, if you used irradiance map + lightmap, then saved the irradiance map and used the saved map for further renderings, you can set the secondary engine to None, so that it doesn't compute the lightmap all over again.

When I set lightmap for second bounce, and second bounce multiplier=1.0, everything gets burnt out completely. Values from 0.8 or lower work well, what value did you use in your tests (light grey material images)(how light is light grey )

I think it was (200, 200,200). Both the primary and secondary mutlipliers are 1.0 in my images, since this is the physically correct value. Maybe it will be helpful if I do an animation with the material going from black to white - it's not a linear transition, as one might expect...

I read somewhere that lightmap filtering is better turned off if used for second bounce. I tries various settings, and indeed I only see increased rendertime and no visibile result. When set to first bounce the result of filtering is obvious.

Well, this is true, but without filtering you can get some artifacts. For the large version of the image below, I used world scale for the lightmap, and a fixed filter.

Already on the first two images you have nice AA, area shadows and glossies. What settings did you use to get these results? I tried various combinations of AA image samplers, AA filters, material glossy subdivs, light subdivs and qmc settings, but I never get a complete good result. Either the area shadow doesn't look good, or the AA on the window frame gets jagged, etc... I'm especially confused about the min samples value of qmc, and about the adaptive qmc AA. With adaptive qmc, values of 4/4 for example give me better results than 4/6 or even 6/8. But using equal numbers here is not what adaptive is used for of course...

AA was default Adaptive QMC, QMC sampler was default (this was changed for the later renderings though), and glossies and the light (I have only 1 area light at the window) all have 30 subdivs.

About touching geometry, I know this is a problem for every GI renderer or radiosity solution etc. But Problem is nobody wants to model a product 2 times, one for rendering with overlapping and one for production without overlap. Don't know about archviz but usually when you get a building model, it isn't modeled with overlap either, so you don't want to remodel it of course. The check sample visibility option does a great job already on this matter!

Yep, actually I don't think it was necessary to move the floor; I did it just in case...

Good to know that IR map will take longer with bright multipliers, I didn't think of it that way. What values are considered really too high (as in, what should you really avoid?) 6-8-10?

That depends, but the number of samples required for a clean result is typically proportional to the brightness, so if you make your lights twice as bright it is likely that they will require twice as many samples to get the image right in terms of noise.

Larger rendering; there are still some artifacts to fix tomorrow, but that's it, more or less:
http://www.vrayrender.com/stuff/flipside/test7.jpg


Best regards,
Vlado

[vlado]

And the gamma-corrected version:



Best regards,
Vlado

[TRGraphics]

What do you mean by "gamma correction". Do you use photoshop or some other program?

Tony

[vlado]

What do you mean by "gamma correction". Do you use photoshop or some other program?

For the big image I used HDRShop (saved the image as a .hdr file first); sometimes I just use IrfanView with a .png, as with the smaller image above.

Best regards,
Vlado

[flipside]

Hi,

To compare rendertimes a bit, could you post the machine specs also?

If I set the AA and qmc to default, I get more noise than in your tests:




Any ideas what is wrong?

My settings:


Regards,

wouter


[edit] ok, probabely the default qmc of 1.46.05 is noise th=0.005 instead of 0.01 [/edit]

[TRGraphics]

I've tried vlados settings and I still can't get clean glossy reflection.

I would like to have the scene he set up to see what I'm doing wrong.

I've tried everything I can think of, and they still look like yours Flip.

Perhaps the build he's using fixed the problem?

Oh, when I saved the lightmap using his settings, it was 547 meg.

Tony

[Long John]

For the large version of the image below, I used world scale for the lightmap, and a fixed filter.

Best regards,
Vlado

why world scale?

in the help file is written that screen is better for stills.

[Sawyer]

Flipside or Vlado if either of you could post your lightmap settings I know I would appreciate it. I have not gotten lightmap to be anything but very slow.

Thanks

[cocolas]

yeah, I'd like the scene as well, vlado.


this was a great idea btw, wouter....




paul.

[paulison]

nice post,

please show the 'adjusted' scene vlado (if small enough), i think it will answer most peoples questions

thanks!

[vlado]

Of course I'll post it, I just want to make a few more tweaks and make something like a short tutorial about it.

Best regards,
Vlado

[pengo]

excuse me VLADO maybe (surely) mine is a stupid question but...

Why u did a gi calculation of white solution first of all?

to mix with the gi solution with final materials applied to get brighter results and desaturate results?
has something to do with imap viewer mixing the 2 gi solutions (the first with white materials and the second with all amterials applied)?
or to test simply the scene to see better the artifacts?

thanks for your support (obviously GREAT support)

excuse my BAD english
bye

[vlado]

or to test simply the scene to see better the artifacts?

This is the one reason. The other is that I wanted to adjust the lighting first, and it's easier (and faster) without all the glossies and bitmaps getting in the way.

Best regards,
Vlado

[pengo]

thanks VLADO

bye