Lock FGMap and PhotonMap to Camera View

[ivanjay]

I often create multiple interior camera views of the same scene. Two questions:

 

1. Do I need to individually create photon and final gather map files for each camera view(I am assuming yes for final gather map and no for photon map)

 

2. Is there a way to attach a specific map file to each camera view so i do not have to change it every time (this becomes cumbersome with rendering farms and leads to errors)

 

Thanks!!

[Justin Hunt]

If nothing changes in each view, like light settings or object moving, then you can have one of each Photon and FG map. Think of it as an animation that steps through each view.

 

Just leave the read/write to file option and calculate each view.

Have "Calculate FG/GI and Skip Final Rendering" ON

Then lock your Photon and FG maps and render all the views

 

jhv

[ivanjay]

If nothing changes in each view, like light settings or object moving, then you can have one of each Photon and FG map. Think of it as an animation that steps through each view.

 

Just leave the read/write to file option and calculate each view.

Have "Calculate FG/GI and Skip Final Rendering" ON

Then lock your Photon and FG maps and render all the views

 

jhv

 

So if I understand this correctly I will use one file but I do need to generate the photon map and final gather map for each camera view... But to confirm, I can write that data to the same file.

[Crazy Homeless Guy]

I am not sure this is true for a Photon map. I thought Photon maps were both geometry and view independent. Meaning, as long as the geometry doesn't change drastically, the Photon map should still be good, and because Photons are traced from the light source, and not from the camera, I would assume that they would be good for different views.

 

Also, you can shoot a handful of cameras that see everywhere in the scene, and have them all add to 1 FG map. Then you should be able to move a camera around the scene, and generate different views. As long as you are not seeing geometry that was not seen by the original camera, then you should be fine.

 

That would work for generating a lot of test views, but for final renderings I would almost always generate new lighting maps.

[ivanjay]

I am not sure this is true for a Photon map. I thought Photon maps were both geometry and view independent. Meaning, as long as the geometry doesn't change drastically, the Photon map should still be good, and because Photons are traced from the light source, and not from the camera, I would assume that they would be good for different views.

 

Also, you can shoot a handful of cameras that see everywhere in the scene, and have them all add to 1 FG map. Then you should be able to move a camera around the scene, and generate different views. As long as you are not seeing geometry that was not seen by the original camera, then you should be fine.

 

That would work for generating a lot of test views, but for final renderings I would almost always generate new lighting maps.

 

I am not really worried about the photon map as those generate in a matter of minutes anyway. It is the final gather map that takes 10 hours or so to create per view.... I was hoping to find a way to speed that up...

[Crazy Homeless Guy]

I am not really worried about the photon map as those generate in a matter of minutes anyway. It is the final gather map that takes 10 hours or so to create per view.... I was hoping to find a way to speed that up...

 

In that case, you might want to post your scene and computer specs. This is an abnormally long time. You probably could get by with draft settings. The higher settings are more for product shots than they are for arch viz shots.

[ivanjay]

In that case, you might want to post your scene and computer specs. This is an abnormally long time. You probably could get by with draft settings. The higher settings are more for product shots than they are for arch viz shots.

 

Oh boy...

 

Quad Core AMD 2.5 ghz, 4 gigs RAM

Decent video card but I know that really doesnt matter

64 bit OS.

 

I can send a zip of a model my colleague is working on if you pm me your email address. He is still learning so there are a few issues in there I am reviewing but it gives you an idea of the size of the model. This one is only one station, sometimes we do an entire cafeteria which might be roughly 2-4,000 square feet depending on the size of the space...

 

I normally pull the slider to medium for the final gather map file creation. I used to do it at full resolution but have read about 1/2 size and recently started doing that. I do not notice any quality issues so that is helping tremendously.

[Crazy Homeless Guy]

I am not sure I will be able to get to it tonight. Working on a deadline right now. Maybe Justin has a few minutes.

[ivanjay]

I am not sure I will be able to get to it tonight. Working on a deadline right now. Maybe Justin has a few minutes.

 

Doesn't need to be tonight. This is a topic for future knowledge for me, not on an active project.

[Justin Hunt]

Photon maps are global, ie not view dependent. So only has to be generated once.

 

FG maps are view dependent. However they have the ability to interpolate between points generated from different camera locations (As what happens in an animation and FG is calculated at every nth frame)

 

Draft should be fine, medium has more detail and takes longer to generate. Go with what works for you in terms of time and detail.

 

10 hours is way out.

 

Post a screen grab of your settings and once of a typical scene. Be sure to unhide the lights as well.

 

jhv

[ivanjay]

Attached are my settings I typically use...

[Justin Hunt]

A few things jump out

 

1) there are alot of lights

2)the FG desity of 0.8 if quite high when using GI, Draft with a higher interpolation will do

3) The Global Tuning parameters are quite high. Rather control such settings on a per material level as needed.

 

How many Photons are you shooting?

 

If you'd like, can you shoot me the file, I suspect there are other issues. PM me if your interested

 

jhv

[ivanjay]

A few things jump out

 

1) there are alot of lights

2)the FG desity of 0.8 if quite high when using GI, Draft with a higher interpolation will do

3) The Global Tuning parameters are quite high. Rather control such settings on a per material level as needed.

 

How many Photons are you shooting?

 

If you'd like, can you shoot me the file, I suspect there are other issues. PM me if your interested

 

jhv

 

Couple of answers:

 

1. There are a lot of lights. And this is a relatively small model. However, it is accurate to what is really going to be there. We have general lighting in the ceiling, lighting in the equipment, and accent lighting on almost every project.

 

2. I have no idea what settings are truly necessary, hence I just move the slider. What would you recommend putting density to?

 

3. I generally do crank up the per material settings for global tuning as well. I have seen that I normally need to set the global reflection to 2, this model was a bit different because there are a lot of reflective surfaces.

 

Can you pm me your email address and I will send you the model?

[mattclinch]

i've never had to go above 1 for the global tuning values. those values kill FG/AA times...

[Justin Hunt]

sent you a PM

[Justin Hunt]

I have had a quick look and can track down a few issues

 

I had a look at the model over the weekend and these are some of the things I picked up.

 

1) Displacement took a long time ( problably dont need it and it wasn't set up optimally)

2) Set FG to Draft really sped things up , Just with these two I almost halved the rendertime

3) Photons, I set a photon radius and turned on "All objects generate GI"

4) Set all Glass objects to "Pass Through" Final Gather and not to generate GI, under the object properties-> Mentalray tab

5) Set near and Far attenuation for all lights, this helpes quite alot

6) As a gerneral guide try to use A&D shaders for everything. Personally I dont like the Pro-materials as they dont have enough control

7) mentalray doesn't like shadow maps, use raytraced shadows instead.

 

Generally things weren't set up wrong, they just needed a bit of tweaking.

 

There are more tweaks that can be made, mainly with the materials, I didn't really play with those much

 

sent the model back with my tweaks.

 

HTH

 

JHV

[ivanjay]

I have had a quick look and can track down a few issues

 

I had a look at the model over the weekend and these are some of the things I picked up.

 

1) Displacement took a long time ( problably dont need it and it wasn't set up optimally)

2) Set FG to Draft really sped things up , Just with these two I almost halved the rendertime

3) Photons, I set a photon radius and turned on "All objects generate GI"

4) Set all Glass objects to "Pass Through" Final Gather and not to generate GI, under the object properties-> Mentalray tab

5) Set near and Far attenuation for all lights, this helpes quite alot

6) As a gerneral guide try to use A&D shaders for everything. Personally I dont like the Pro-materials as they dont have enough control

7) mentalray doesn't like shadow maps, use raytraced shadows instead.

 

Generally things weren't set up wrong, they just needed a bit of tweaking.

 

There are more tweaks that can be made, mainly with the materials, I didn't really play with those much

 

sent the model back with my tweaks.

 

HTH

 

JHV

 

A few questions / comments:

 

1. We use the displacement for the food to give it some 3d texturing. You mentioned it "probably was not set up correctly." Can you expand on that so I can ensure I tweak that to be better in the future?

2. If I make it draft isnt that a really low quality setting for production renderings? I do that for previews, but I put on medium for production.

3. I did some reading and will start to play with photon radii. It sounds like it is a better idea to size the photons and play with the quantity and preview rather than setting a blind setting as I previously did. How do I set up to view the photon map? I have seen that done elsewhere. I checked enable under diagnostics and selected density and photon but it is rendering my image.

4. Why would I want to pass through for glass in final gather? Wouldnt that produce inaccurate results or will I not really see it?

 

I will try everything else.

[Justin Hunt]

Displacment in mentalray is relyant on a dense mesh, the object used as food were not dense enough. Use a A&D shader, down near the bottom is the mentalray connection tab, unlock the displacment and drop a Height Map in there. then control the displacment through the render dialogue.

 

However considering where the food is placed in the view, (Mid-ground and behind glass) how much of the displacment effect you are really seeing, and is it worth the render/ memory overhead?

 

FG, when using Photons, FG is used more to clean up or smooth out the photon solution. Photons calculate quicker so let photons do the hard work by increasing their settings before relying of FG to add detail.

 

Photon radius is very important, if left unchecked mentalray will decide what is best. Generally is 1/10 of the scene bounds, which is fine if the area of interest in the view is close to the scene bounds, but not so good if the scene bounds is much bigger than what is being seen in view. There is another thread http://forums.cgarchitect.com/40576-gi-question-radiuses.html, where is discuss this in a bit of detail.

 

I like to set glass to Pass through mainly because the visible effect glass is having in FG is neglegable, but the impact it has on the calculation and render time can be significant. I prefer the FG samples to work on areas that will have impact. This way I get more FG samples where they are needed, I can use fewer samples, and quicker render times. Of cause if your doing stained glass windows it is better not to set it to pass through.

 

If accuracy was really that much of an issue then you need to pay more attention to your material and lighting setup.

 

jhv

[ivanjay]

Displacment in mentalray is relyant on a dense mesh, the object used as food were not dense enough. Use a A&D shader, down near the bottom is the mentalray connection tab, unlock the displacment and drop a Height Map in there. then control the displacment through the render dialogue.

 

However considering where the food is placed in the view, (Mid-ground and behind glass) how much of the displacment effect you are really seeing, and is it worth the render/ memory overhead?

 

FG, when using Photons, FG is used more to clean up or smooth out the photon solution. Photons calculate quicker so let photons do the hard work by increasing their settings before relying of FG to add detail.

 

Photon radius is very important, if left unchecked mentalray will decide what is best. Generally is 1/10 of the scene bounds, which is fine if the area of interest in the view is close to the scene bounds, but not so good if the scene bounds is much bigger than what is being seen in view. There is another thread http://forums.cgarchitect.com/40576-gi-question-radiuses.html, where is discuss this in a bit of detail.

 

I like to set glass to Pass through mainly because the visible effect glass is having in FG is neglegable, but the impact it has on the calculation and render time can be significant. I prefer the FG samples to work on areas that will have impact. This way I get more FG samples where they are needed, I can use fewer samples, and quicker render times. Of cause if your doing stained glass windows it is better not to set it to pass through.

 

If accuracy was really that much of an issue then you need to pay more attention to your material and lighting setup.

 

jhv

 

I cannot thank you enough for the timing to put together such a great response. I would consider myself intermediate at best. I have good knowledge of the basics but am working to elevating my "game" a bit. This detail is soooo helpful!

 

So, to correct my displacement essentially I just need to pump up the segments for the box I created the food with... That makes sense. I am using displacement as we render for foodservice our clients really are looking to sell their food. And thus it must look good. We have always been stuck with just crude looking splashes of color. Trying to get some depth going will make a big difference. However, modeling the food to the nth degree of detail just doesn't seem to make much sense...

 

The photon comment make sense. I have read through that other thread before but reading it now seems to make more sense. The one thing I cannot seem to figure out is how to preview my photon map creation. I checked diagnostics under processing and do not see any change....

 

The glass comment makes sense...

 

My comments about realism are not that I truly need this to be 100% accurate. But, am looking for the non cg artist to take a look at it and feel it looks pretty realistic...

[Justin Hunt]

To see the photon size , turn FG OFF and you will see circles of light, this is the photon radius.

 

I like to know/ understand what the settings mean so I find this process as helpful to me as I hope they are for you.

 

jhv