Motion Blur & Depth of Field: Done in Post?

[Claudio Branch]

Is there any advantage to doing Motion Blur and Depth of Field in a compositing program vs. directly out of the render engine?

 

I am curious if there is more control over the quality and less of a hit to render times by taking this approach...

 

I have had decent results using Motion Blur, but I have never been satified with the results of depth of field.

[BrianKitts]

definitely post..... I don't have any experience with motion blur, but with depth of field when you save out your zdepth channel and apply the DOF in post you can control the amount applied to the image. If you do it in the render it gets "baked" in and you can't adjust it after the fact. Which would mean rerendering again if it doesn't come out right.

[dinhtuan]

It's huge rendering time if we use montion blur and DOF of render engine . IMHO Reelsmart Montionblur is good and some blur with zdepth channel is ok for DOF like BKittsARC said. Did you try Depth of Field Generator? there're batch section very usefull

[Claudio Branch]

 

Did you try Depth of Field Generator? there're batch section very usefull

 

 

I do not know what this is. Could you elaborate please?

 

My understanding is that I can render out a depth map from Max and then use that within After Effects.

 

If the camera is moving, then wouldn't it require a z-map for every frame as well?

 

I am beginning my compositing career with this post I suppose...

[Chad Warner]

If you render to an rpf file type, you can put the z depth information as part of the file that after effects reads. The other alternative is rendering a max element that contains the z depth info and using that as a mask in after effects.

[dinhtuan]

Depth of Field Generator: http://www.dofpro.com/

With camera moving, we still need depth map for eveyframe, after that we use them as mask effect for DOF in Photoshop. with Batch action we can apply for sequence images of the animation.

 

http://www.dofpro.com/batchgallery.htm

Hope this help

[Claudio Branch]

That was pretty helpful!

[Crazy Homeless Guy]

aftereffects is rather lackluster in its handling of z-depth, but i think there are plug-ins for it that do a more elegant job. basically, you can only blur at about .5 or .75 before it starts to look odd where the background hits something in the foreground. severe artifacts and such.

 

you have accuracy when rendering DOF in the 3dsmax using a physical camera and adjusting the camera settings like you would in real life, but it comes at the cost of being permanent, and drastic increases in render times.

[Claudio Branch]

I appreciate the input regarding After Effects...anthing to help reduce the learning curve! The stuff from DOFPro looks very strong and the full-version demo is rather appealing.

 

I can probably stomach longer rendering times for stills...especially using distributed bucket rendering. But most of the stuff I do is animation, so I want to establish my workflow with that in mind.

[Limbus]

Both methods have their advantages and disadvantages.

DOF in post only works when you don't have transparent Objects like glass tables in your scene. The transparent object will turn up solid in the z-buffer and therefore a post DOF effect will render everything behind the transparent object w/o any blur. Rendered DOF does not suffer from this but you cant adjust it afterwards so if an Art Director wants some changes you have to rerender. I try to use post DOF if possible.

 

To get the best results with MB in post you really have to split up your scene in multiple passes to seperate the moving objects from the static background and render out a motionvector file. This is explained quite good here:

http://www.revisionfx.com/support/faqs/generalfaqs/motion_vector_alpha/

http://www.revisionfx.com/support/faqs/generalfaqs/motion_vector/

http://www.revisionfx.com/support/faqs/generalfaqs/motion_vectors/

 

You can also do a post MB effect w/o a motionvector pass but the quality will be lower.

 

Both, DOF and MB done in post suffer from the limitation that real DOF and MB might reveal parts of the image that are occluded in a normal render. Obviously a post effect cant mimic that. This is explained here:

http://www.frischluft.com/lenscare/lenscare.php (at the bottom)

 

So renderd DOF and MB has better quality but renders longer and cant be adjusted afterwards.

 

Florian

[Claudio Branch]

It's huge rendering time if we use montion blur and DOF of render engine . IMHO Reelsmart Montionblur is good and some blur with zdepth channel is ok for DOF like BKittsARC said. Did you try Depth of Field Generator? there're batch section very usefull

 

Is Premiere Pro required for use with these? I am using Premiere at present.

 

Also, can you describe the workflow involved using DOFPro?

 

My understanding is that I create a z-depth map within Max, use that map within DOFPro for creating the depth effects and then loading the original frames and the corresponding DOF filters into Premiere or AfterEffects?

[John Dollus]

Just chiming in to backup florian - motion vectors with Reelsmart plugin will give you very nice results

[Tommy L]

I have found using AfterEffects for depth of field can be fiddley, but I am still using AE 6.5. It may have got better. Ive heard combustion handles it much better or as already stated, plugins for AE maybe the way forward.

 

The reason DOF and Motion Blur in the render engine take so long to render is because the render engine renders the blurred parts of the scene multiple times then composites them.

 

I havnt tried doing the batch render technique in Photoshop for this though, nice and controllable.

 

The main things to remember are: get the z-depth pass right and dont over do it!