MR: strategies for faster FG

[nisus]

Hi all,

 

What strategies do you use to reduce rendering time when using MR with FG?

 

Imho one can save some time when one does not have to rebuild the fgm-map over and over again.

BUT, if using a fgm-map speeds up rendering time, what is the best way to use it?

- Render a fgm-map at full res with an ideal diffuse material override?

- Render a fgm-map at low-res but with very high FG settings?

- something else?

 

Any tips are welcome, a discussion would be great ,-)

 

rgds

 

nisus

[William Alexander]

Nisus,

 

do you mean, for clarity, reusing the FGM for iterations of rendered scene improvements?

 

Building a render solution generally finding that, it depends on what areas your focusing on.

 

Macro (overall conditions), low res higher FG settings

GI affect, High res, low to med FG settings

Detial materials & GI, med res, med FG settings, use render region for specifics.

 

Generally, start low res, low fg setting, rebuild, with about a 2 minute max render time including FG. Then as the scene develops use the approaches mentioned above to further refine. Also use the AA settings to improve rendering speed, can knock off some significant time.

Keep the FG settings low, res med, untill you need final renders-Then Incremental increases; 50%-100% res, then raise FG (the scene gets brighter than low FG sample settings, BTW), then raise AA to high levels.

 

It's a juggling act of variables. Takes a little time to get the feel for what's going to do what. Every scene tends to have different variables to attack. Try a couple of test runs if using the radius diameters early on full res, using the material override. Getting those settings right early will be an aggragate time saver, then use default to develop the solution. Come back to the settings for the final verification renders.

 

Damn, another long post. As you can see, using as many of the controls as possible yields the best results, just not a clearly defined method. It's like a complex flow chart using conditions to determine actions.

 

Cheers

WDA

[joske]

both interior and exterior arch scenes should render ok with very low FG settings

so to put it simple and blunt : to save time use LOW FG settings all the time

by low I mean not more then 500 (25-50 for previews 250-500 for final result)

 

if the result does not look good... your photons are not setup ok, or the MR light samples are to low (or your scene and/or max units are not 'real size')

 

My workflow is to turn off FG to start and see what the photons are doing. make sure they look well distributed and evenly spread and the intesity looks real

 

then put on a little FG to test and you should allmost have final result, if you see some artifacts then turn up FG slightly until gone.

 

imexp using LOW FG settings is the key for fast preview and final results in Mray

but in order to use low FG setting your photons should be 'spot on' and setup ok.

less, and bigger photons will render faster but requier more FG

more and smaller photons will render slower but requier less FG

 

it's just a matter to find a balance with the fastest render that's the hard part

you will find what works best for you by just doing scene after scene

(my way is not 'the way' but it works for me...)

 

last thing i might add : every scene requiers a different setup so importing another building into a scene that was made before does give you a good starting setup, but will allmost allways requier some tweeking to look perfect.

[kicks]

- Render a fgm-map at full res with an ideal diffuse material override?

nisus

 

[nisus]

Hi kicks,

 

I haven't tried it yet, because I just didn't havet he time to experiment yet... I was just asking so I could reduce the things I wanted to try out for now.

 

Hi WDA and Joske,

 

Tnx. Your tips go about the same line as what I do for now, but I still think it's to slow. I try to fasten up things by setting up lighting levels first (with an ideal diffuse material override) and coloring/texturing afterwards so I don't have to worry about settings when texturing.

 

For the GI, I ONLY use FG, so without any photons. Somehow I find that things go to slow when shooting particles, or the result is too horrible when shooting not enought...

Do you have any tips on using the GI and FG? I mean, what advantages are there using them together, or one or the other? For now, I don't have any goodlooking results with the GI an photons...

 

Is it possible to post a serie of testrenders and settings of the seperate steps that you use in your workflow? Tnx

 

rgds

 

nisus

[joske]

Do you have any tips on using the GI and FG? I mean, what advantages are there using them together, or one or the other? For now, I don't have any goodlooking results with the GI an photons...

 

rgds

 

nisus

 

imexp / workflow it is very hard to get good results only using Fg :

 

1. your ambient light in the shadow side of the scene will only come from the skylight, and not from the sunlight/photons (i use a Mr spot for sun)

when using photons you will add ambient light from the spotlight (like in real life)

 

2. using photons will allow you to use less Fg and using Fg will allow you to use less photons :

so another balance to consider to achieve the best vs. fastest result.

 

a bit of both worlds would give the best and fastest result imexp.

if one is way bigger than the other, you will have longer rendertimes with similar results in most cases.

[nisus]

Hi all,

 

Does anyone knows whether the fgm-map is resolution independent or not?

 

In other words, can I render a fgm-map at 180x360 and use it for a 3800x7600 image without losing quality? Or does it has to be a rendered at high-res too?

 

to kicks: I am running the tests with a fgm-map and an ideal diffuse material override (*IDMO) at the moment. I'll keep you informed about the results.

Reusing the fgm-map with the same IDMO-scene got rendering times from 08:07 to 01:14. Gonna check it out when using the same IDMO-fgm in the normal scene...

 

rgds

 

nisus

 

 

 

*IDMO: ideal diffuse material override

[joske]

Hi all,

 

Does anyone knows whether the fgm-map is resolution independent or not?

 

 

Fgather is calculated from the camera viewpoint

(Photons are calculated from lights)

 

so I would assume it is resolution dependent, however that said it would be not a bad idea to still do some tests... i'm not sure at all

 

I tested the fgm map (use file) but it still is a mistery when to use this.

1. Because it almost allways recalculates anyways, and 2.because it is camera view dependent...

so i don't really see the benefit or time gain in like animation renderings or so ??