View Full Version : FG adding additional points in 2nd pass
subaqua
November 25th, 2005, 09:24
I was wondering if someone can help with one of the most frustrating issues I’ve encounter with Mental Ray.
I’ve rendered out a FG pass and when I freeze it and go to render the second main pass, MR keeps saying something like this as it is being rendered:
1000 additional points for the tile [128 191] x [256 319]
2000 additional points for the tile [156 194] x [263 343]
The frame will render but it takes ages. It seems to be redoing the FG even though the FG has already been calculated and frozen.
Can someone please help with this issue?
I’m using Maya 7 on Windows XP SP2 by the way.
bart
November 29th, 2005, 16:23
Are you saying that this happens at render-time as opposed to FG pre-processing time. In other words, if you do an FG diagnostic, do these show up as red points?
What are you changing between the FG pass and the second pass?
Resolution?
In standalone, there is a deepfreeze mode which guarantees no FG points will be added during render-time.
-finalgather_rebuild d
subaqua
November 30th, 2005, 08:01
Yep at render time it does it. So it will render a bit then add more FG points then render a bit more and will continue to do this.
Since the FG pass nothing has change in the scene, so there should be no need for more FG calculation. Plus with Freeze FG you think it wouldnt add anymore points.
That deepfreeze option sounds good, I whish it was in Maya 7.
It seems that Mental Ray in Maya 7 is a bit buggy. I never had any real issues in Maya 6.5
bart
November 30th, 2005, 22:45
FG should be better in 7. I'd like to get to the bottom of this.
What specifically is the bug you see?
Can you outline what you are doing?
For example, are you using the FG only render option for an FG pass?
Do your scene objects have lots of surfaces that are like table tops viewed from the side? That is a typical cause for render time FG points to be calculated, because the view-dependent scene pre-processing doesn't turn around corners all the way. However, the version in 7 should have better FG point adaptivity. Maybe you could increase FG presample density.
I'd like to see your FG diagnostic image.
subaqua
December 2nd, 2005, 08:46
My main issue is that I don’t want anymore FG points being added during render time, as it is making my renders take way to long. I thought Freeze FG stopped this all round, but it only seems to stop it adding to file.
First I go into “mentalrayGlobals” and selecting “Render FG Map”, so I can pre calculate all the FG info so it doesn’t have to during render time.
Here is a final render of my scene http://www.deviantart.com/view/25879841/
I’ve done two renders one with FG in Pixel size and the other in object space.
The render in Pixel size seems to want to add more FG points at render time.
What I’ve found is the higher my AA sample setting the more FG points get adding at render time. And FG in Pixel size adds a lot more then in object space.
Here are the FG diagnostic images. (note) don’t worry about the red in the tree on the right, as I didn’t render it in the FG pass.
FG in object space:
http://www.deviantart.com/view/25912400/
FG in Pixel size & AA sample settings of -1, 1. Contrast threshold of 0.1. Render time was 18min
http://www.deviantart.com/view/25912468/
FG in Pixel size & AA sample settings of 0, 2. Contrast threshold of 0.03. Render time was 58min
http://www.deviantart.com/view/25912516/
You can see a lot of red going into the higher AA render.
So all I want is for MR to not bother trying to add anymore FG point to the scene even if there is a low FG ray count.
You said about a deepfreeze option for Mental Ray standalone. So would this do what I’m asking? If so it would be really handy it get put in to the next update of Maya.
Also would you be able to direct me to who I can contact to submit requests for MR options in Maya.
Thanks
Leo
bart
December 2nd, 2005, 16:34
Please submit your requests to Alias.
If you don't change any FG parameters, but increase the number of samples, you will have more chance of getting red (render-time) points as you have noticed in your last picture. I suggest increasing the FG points in your FG map pass. Since you are reusing it anyway, the extra time creating it will be worth it.
Did you try increasing your FG preprocessing sample point density?
(See extra attributes of mentalrayGlobals or miDefaultOptions node.)
You could also decrease FG max radius, but you would get less FG points interpolated for a given fixed area. Less soft.
subaqua
December 3rd, 2005, 08:49
Do you know which email address to use to contact Alias for a request?
By increasing FG preprocessing sample point density, do you mean up the number of FG rays? In my scene I used 5000 rays. I tried decreasing the FG max radius but the lower I went the more FG points go added at render time. I don’t seem to get any red points when I have a larger Max radius, but then I get allot of blotches around the glass.
Any who I guess all will be solved once that deepfreeze option gets added into Maya.
bart
December 4th, 2005, 03:30
No. I mean a parameter that is only exposed in extra attributes of mentalrayGlobals or miDefaultOptions node.
You should post the solution you found.
I can summarize, but you should add more details.
Use smaller FG radii for creating the FG map, then back for using it.
That is similar to increasing presample density, but maybe more effective for your particular scene. I'd like to compare the approaches.
Also, a similar brute force manual approach is to append FG points to the FG map file, by zooming in on the red areas, and maybe tilting the camera a bit to get around the deep corners which cause the red in the first place.
[BTW, you append by using FG rebuild off.]
subaqua
December 4th, 2005, 05:00
Could you send me a screen shot of the option in mentalrayGlobals for FG presampling, coz I can’t seem to find it.
I tried zooming into areas and then zoom back, but that ended up being to time consuming on a large scene as I had to going into every spot then wait for it to render before I can go to the next spot. But yes the end result was similar.
leha_sokol
December 7th, 2005, 08:29
use this mel command
select miDefaultOptions;
Then go to extra attributes section and there you find this FG Presample denstity attribute...
subaqua
December 11th, 2005, 02:02
wow i never knew that bit was there. Thanks heaps for the m8. so how is it use? what values do what?
Also do you know what the Hardware options are?
leha_sokol
December 12th, 2005, 11:23
Well i m far from being an advanced MR user ....so dont know much about all MR trick and hidden attributes...just dig a little on this forums ...there are heaps of intersting tricks here to read...
From what i know (besides Presample Density) there is an option for calculation FGmaps alone without actual rendering the whole image...take a look into Rendering tab of miDefaultOptions
Cheers mate :wink:
yashu
December 23rd, 2005, 09:11
i'm also getting loads of extra FG points added *on render time* in maya 7 and i heard XSI 5 has this prob aswell, so it must be a bug in mentalray 3.4.5 or higher cuz in maya 6.5 FG + FG multibounces was giving a *much* clean result in hell less time.
i guess there isn't any fix for this yet, so we must wait for mr 3.5 :|
subaqua
December 23rd, 2005, 09:52
i'm also getting loads of extra FG points added *on render time* in maya 7 and i heard XSI 5 has this prob aswell, so it must be a bug in mentalray 3.4.5 or higher cuz in maya 6.5 FG + FG multibounces was giving a *much* clean result in hell less time.
i guess there isn't any fix for this yet, so we must wait for mr 3.5 :|
There is actually something you can do for it. But it only works if you’re doing your FG in a pass. What you do is, do all your FG calculations then freeze the FG file. Then you up the Min, Max radius for the FG when you’re ready to do the final render. Works like a charm.
But yea I was never having this issue with Maya 6.5
yashu
December 23rd, 2005, 10:41
There is actually something you can do for it. But it only works if you’re doing your FG in a pass. What you do is, do all your FG calculations then freeze the FG file. Then you up the Min, Max radius for the FG when you’re ready to do the final render. Works like a charm.
But yea I was never having this issue with Maya 6.5
so you mean i have to render the FG maps at full quality first and then freeze them and render again but with lower quality FG? did i get it right?
thanx a lot :wink:
Merry Cristmas!
subaqua
December 23rd, 2005, 13:34
Yep thats right. So go into the “select mentalrayGlobals” and choose to render FG map only. You can also try increase the FG Presample denstity.
Ziko
January 18th, 2006, 17:31
Hi,
I got the same problem, I tried your solution but it dont work.
When I did a FG test with a lambertian shader to fine tune my fg and I got the good picture from diagnostic everything is ok.
But when I put my final shader something goes wrong.
I have also this message, "API 0.0 warn 302018: type conflict: "vector" is type vector" I dont know if there is a relation between this error and the adding purpose.
If someone got a solution It would be great.
I have increase my presample density and freeze the map but it dont work.
The strange thing is that it add points at the start and at the end and its not continue.
I really dont know whats going on.
bart
January 19th, 2006, 18:35
FG points are created in a view dependent fashion.
Is your camera locked in the same position?
So you previously constructed the FG map with higher density FG points and then turned the density back down when you used the FG map?
You can also increase FG point density by zooming in the camera in the trouble spots and then bringing the camera back to the original position.
You could also decrease the FG max radius to increae the FG point density.
All these techniques may bring different results to the final 'softness' of the FG look.
Keep in mind that the FG max radius acts like a filter size control for illumination. It is the radius used for collecting FG points to average. A higher density averages more FG points. But the radius itself controls how far across the object(s) you will be averaging.
One final point. You could increase the FG min radius up to max radius to decrease the number of 'red' render-time FG points. But this might make light leaking occur because of the FG point validity heuristic. Every FG point within the FG min radius is used, unless obviously on the face of another object facing a 180 degrees the other direction. That said, I think a lot of scenes could get away with a lower ratio than the 10:1 max to min recommendation.
The key to understanding how the radii affect your image quality is by checking out the lighting in areas of surface detail.
Ziko
January 19th, 2006, 19:11
Thx for the answer.
I did all the solution you saide before except zooming on the area.
But I dont have enough this problem, I have chance my shaders and it works.
On some object a shader with a high reflection cause this bug I think.
Thats why I thought the warning on the vector was my problem.
But I dont know how to fix it.
But anyway thanks for your answer.
P.S: My radii ratio is 3,3:1 and I got an almost very good diagnostic I think ^^.
subaqua
January 20th, 2006, 01:54
Hi Ziko,
When you’re creating your FG map are you also including what the object reflection is seeing or just what the camera is as if there was no reflection?
If you have a large area which is reflective make sure what is seen in the reflection is covered in your FG map as well.
Ziko
January 20th, 2006, 11:11
Its an interior scene with a floor who is reflecting.
The only one object who is not totally covered by the map is the roof and the back wall.
But I am using the mib_glossy_reflection and I set a max_distance to reduce the reflection.
But I will try to take multiple shot from different view for the Fg map and see if the problem will be resolve.
Thx
bart
January 21st, 2006, 17:27
Hey subaqua,
Good point.
Specular and glossy rays may reflect to diffuse areas which need good FG coverage. This can get tricky and visualizing your light paths helps solve these type of problems.
Also, Ziko, in mib_glossy_reflection, could you list some of the parameters, such as number of samples (glossy reflection rays)? Spread? Etc.
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.