View Full Version : 3D Film/Video Work Flow - Quality Concern

Digital Hermit
05-14-2006, 02:19 AM
Here are some questions for the film and video production experts out there. I would like to know a good workflow process and common conventions for the following apps I have.

Lightwave 8.5
Adobe Premier 1.5
Adobe After Effects 6.5

My main concern, in all of this, is time vs loss of quality of the original file.

I am learning to work with 3D animation, some 3D-to-3D compositing work, and some 3D-to-Live compositing work.

My questions and concerns are:

I render my 3D animations to a "TGA" file format. (Is this the best file for Adobe?)

I render antialiased frames, but I do "motion blur" in AE. (Is that advisable or should I do it in LW or some other process? ... I think doing it in AE saves time... etc.)

I do most post work in AE, with a least the minimum effect of "motion blur" applied and save it in a "file" format or as an DV/AVI. (Is it advisable to save the "AE modified" animation files in the same file format? In other words, I imported it as a TGA file, should the post-worked file be saved as a TGA file? Also, saving the files as a DV/AVI is that better than file format?)

I edit and complete in Premier, my sequenced files and/or my DV/AVI files. (Is there some quality caveat or different process I should be aware of when doing this?)

Final question... I am having problems with Alpha Channels when it comes to compositing in Adobe... mainly transparent objects like windows, etc... (Is Alpha Channel compositing ok, or is there a better process with the apps I have?)

Thanx, in advance for your help,


Digital Hermit.

Digital Hermit
05-14-2006, 02:27 AM
Wanted to delete this... it was a double post... but all I could do is edit. - heh

05-14-2006, 03:29 AM
Work with the individual frames and then convert the finished frames to a movie format last.

05-15-2006, 03:56 AM

I tend to render everything out to 32-bit Photoshop (PSD) sequences to beat up in AE, but that's just habit really, TGA-32 should do the same job. If I know I'm going to be happy with LightWave's blur, then I'll use Dithered motion blur - especially for things weaving their way across the frame. AE blurs are very quick compared to LW's, but they're not necessarily the best. Depends entirely on what I want to acheive with the shot.

Working constantly to uncompressed 32-bit image sequences eats disk space like so many electronic hogs, but compressing to any codec - and DV is a compressed codec (codec is just short for "COmpressor DECompressor in case you're wondering) - is lossy. It's fine to compress once for delivery, but otherwise you're un-necessarily losing information from your frames.

With regards to alpha channels, there's a specific control for that in the Advanced tab of your surfacing options, which gives you surface-by-surface control over your alpha levels. However, if you're saving out to TGA-24 rather than TGA-32, then the alpha data won't be saved.

You can also save your alpha data as a separate image sequence, which can add flexibility without having to re-render.

Hope that helps!

05-15-2006, 04:12 AM
Hi Digital Hermit,
One piece of advice I can give is always render output at medium AA in Lightwave (or above) or the equivalent of.

Trust me I worked that one out the hard way 1000 rendered frames that look poor when output to TV was a real lesson learned.

Other advice is always composite your scenes and render eliments

E.G you have a character in a scene do one render of the virtual set(back ground) then another render of the character, the case being you have to make changes to one or the other render times are minimal as your only rerendering what needs to be changed.

05-15-2006, 05:27 AM
Agreed, stay away from codecs (especially DV!!) for as long as you can (till last step. if possiable) as each generation causes more damage.

Casey :cat:

Digital Hermit
05-15-2006, 07:17 PM
Thanx for replying everyone.

SG and Bog... I want to be clear on the individual frames aspect... So, if I load my sequenced frames in AE and modify them, I should then save them as sequenced individual frames, until I need to "compress the final" into the desired movie codec?

Bog and Casey - I did not know about DV being a compressed file (makes sense now!) I did notice when I bounced DV tracks around, after layering on composited effects, and saved them in the DV format, there was some degradation.

I use the AE motion blur when doing sweeping a camera pan and/or with very little motion in the shot. But you are right, anything that moving fast in the shot, now, gets the LW treatment. (I did a close missile flyby scene and noticed that LW motion blur was a much better choice. :)

badllarma - So a medium or PLD pass of 9 or greater, correct? Separate elements... hmm... I will have to give a try.


Digital Hermit

05-15-2006, 07:53 PM
Something to think about with the motion blur in AE is Reelsmart Motion blur. It's a commercial plugin (bout 150-200 dollard now days, I think) but it does some nice work. It actally looks at the motion in the scene and only blurs what is moving from frame to frame. Plus you can so some really nice special effects by using the motion from another layer and jacking the effect up (makes nice ghosts) Check it out

Casey :cat:

05-16-2006, 12:47 AM
Thanx for replying everyone.

badllarma - So a medium or PLD pass of 9 or greater, correct? Separate elements... hmm... I will have to give a try.


Digital Hermit

You thats it with regards to a pass, thats on an output of PAL 720 x 576 BTW I don't do any work at higher res than that for TV/DVD. So could not comment on 2K for instance.

Mainly find the seperate elements works when doing Character work more than anything else, before now I've had a client sat on my shoulder wanting a cerain look :argue: and to keep going back and rerendering the entire frames was just not going to happen, not to the point of minimal waiting and saying "is this what you want" anyway.

For all character work I always split the frame up. It helps when getting your timing spot on without the long wait between renders....................... to see if it is spot on!

05-16-2006, 01:02 AM
Yup, render out to uncompressed image sequences. There are some who say that uncompressed .AVI isn't lossy, but I just like image sequences. They're all kinda flexible, easier to handle, and when a client asks for a still, it's right there ;)

05-16-2006, 01:40 PM
I alway render with individual frames, cos if you get an unexpected crash or power failure, you can carry on from where it left off, unlike avi's etc..

I also tend to output RPF's, so I can have the various buffers to hand if I need them.

05-16-2006, 04:36 PM
The reasons to use sequential uncompressed images rather than codecs are legion.

There's only one reason to use a codec: Delivery.

05-16-2006, 05:26 PM
are pngs uncompressed? I know they are lossless but they seem awfully small to be uncompressed.

05-16-2006, 05:32 PM
PNGs work on a similar principle to the old Amiga Hold and Modify ILBM format - staged waypoints through the pallette, followed by an offset, then a repeat value. They seem small because repetative image data is tagged and flagged rather than reiterated. 200 32-bit pixels become a 4-bit flag. It's non-lossy compression.

A rare and special gift from The Culture.

This is where folk like you and I do it Right, and leave the rest in the dust. Smile, because we know something the Muggles don't :)

Digital Hermit
05-16-2006, 09:39 PM
Lottamedia... I checked out the website... the motion blur plugin looks cool, also; they have a few good plug-ins for AE. I might invest in a couple.

Steve, I have to agree with the "failsafe" benefits when rendering to individual frames, that happened to me once and I was so glad that I did not render to avi... Oh yeah, forgive my ignorance but what are "RPF's" ?

Bog, so, it seems from what I have read, that you recommend rendering to PNG's over TGA?


05-17-2006, 01:48 AM

If your comping package uses 'em properly, then go for your life.

Then again, 32-bit PSD can also be very efficient, as again it performs a certain degree of non-lossy compression as far as I can tell. After a certain point, it's a matter of personal preference. PNGs are a great choice for any 16-bit Greyscale work you need to do - transparency, HDR lighting, displacements - or for a seperately-saved alpha channel. If there's any money in using 'em over Photoshop files when using AE, I don't know.

05-17-2006, 04:12 AM
RPF's are format that store extra buffer info, like depth and coverage, vector motion, object ID's etc.. All useful stuff if you a compositing with them. As far as I know, Digital Fusion/DFX, Combustion and After Effects can all make use of these to add things like motion blur and depth of field as a post process, although sometimes the results won't be as good as from LW natively. But it saves alot on render times!

You need to add the extended rpf output filter in the image process tab to save in that format.