Brendan Coots
Forum Replies Created
-
The animation codec is great, but not for video. It’s strength is in encoding graphics and other clips with large patches of solid color, and in the fact that it supports an alpha channel which most 8-bit/10-bit uncompressed formats (such as Aja/Blackmagic) do not. For running video from Avid to AE, you should stick to either the Avid codec OR 8-bit/10-bit uncompressed. If your source footage was shot in DV or something ridiculous use 8-bit, for anything above DVCPRO50 use 10-bit.
-
I watched it without audio and the visuals were really nice. I definitely saw the Video Copilot tutorial influence, but it’s all your work in the end so great job!
-
The #1 problem is that you are shooting DV, which is notoriously difficult to key. the #2 problem is that since DV is so unforgiving, your lighting of the greenscreen itself needs to be spot on, or no amount of post production tweaking will save your shot. The lighting on your actor is important, but nowhere near as important as the lighting on the screen itself since this dictates how much pushing and pulling you will need to do in your keyer app to remove the backdrop, and all that pushing and pulling ultimately degrades the quality of your edges. I typically light my screen with two KinoFlo fluorescent light banks and a 2,000watt Arri light bounced off of a white card on the floor just below or above the screen. here’s a still from a shoot I did last week that, when using the high res source footage, practially keyed itself even though we were only using DVCPRO 50, one small step up from DV:
Your greenscreen should register 50-60 IRE on a waveform monitor (in the image above it is about 53 IRE, I prefer 60). If you don’t have a waveform available (who does), another option is to shoot a quick test clip on set and pull it into Final Cut Pro (maybe on a laptop), which has a reasonably accurate waveform built in.
Also, the lighting must be PAINFULLY even. Variations that are barely noticeable to the naked eye will cause you to push the keyer much harder just to get rid of the entire backdrop. This will cause your edges to get uglier and uglier, from which there is no pleasant workaround.
———-
In answer to your actual question (tangent? Who, me?) here’s a few steps to try that should help your keying:
1. Precompose your greenscreen footage. Within this new comp, create an adjustment layer above the footage, and apply the “Median” effect (Effects > Channel > Median). There is only one value to play with, change it to 3 or so. Put the layer in “Color” blending mode. Now apply Keylight to the composition itself (you will need to nest it within another comp), NOT the source footage. Right off the bat, this will help a LOT of your edge problems.
2. Do not apply any color correction to your plate before doing the greenscreen extraction. Many people try to “cheat” the keyer by boosting values before they key, but it will NOT work and will make your key worse.
3. Break your key up into zones if needed, as this is how the pros do it on feature films. To retain fine hair detail you typically need much more delicate settings than are needed to remove the fringe around sleeves or pants, and one keyer setting applied to both will force you to compromise the quality of both. Duplicate the footage for each area that will need its own treatment, using masks to separate out each element. This way you have one layer that is just the head, one layer for the clothing etc. Then apply different keyer settings to each layer to suit its needs. You will need to animate these masks (depending on what happens in your shots) but again, this is how the pros do it and if you want the best results, this is a good place to start.
-
Not sure about the problem itself, but you should convert the file to .wav, not mov. If your .aiff file is very high quality, it won’t lose anything in the wav conversion.
-
I agree with David 100%. Just say you are still there for them when approvals DO come, but you need to be paid for the work that has already been performed.
Beyond that, this situation should be some cause for concern regarding your future dealings with this company. I am always wary of the print/web/plumbing company that “now does video,” they never know what the hell they are doing.
-
1. RENDER (not export) from After Effects to an Animation Codec Quicktime set to 100% quality, millions of colors. These options can be found in the render queue (once a file is added) by clicking the word “Lossless” in your output module.
2a. Getting your file “to your TV” could mean via DVD or some other method. For DVDs, import the resulting file from step 1 into a DVD authoring application such as DVD Studio Pro (mac) or Adobe Encore (PC) and create a DVD. The program will convert your high resolution Animation codec file into the appropriate format.
2b. If you want to get the file to your tv via some other method besides DVD, you need to tell us what you are trying to do. There are a million ways to go from here.
-
A big advantage of FieldsKit over the VirtualDub implementation is simply options:
– motion estimation-based deinterlacing (which can be good or bad, depending on the situation)
– 16-bit support
– pulldown and reinterlacing options
– Because Fieldskit is a plugin for AE (and FCP as well, I believe) you have more control and don’t have to render a file to get results. Using virtualDub, you would have to import your footage, deinterace, export, work the footage in AE and do a final export. That is a LOT of added rendering and quality loss just to deinterlace your footage.
-
I agree, RE:Vision FieldsKit is going to be better bang for your buck, especially if you only need the deinterlacing features of Magic Bullet. It works VERY well and is attractively priced.
-
All interlaced HD is upper field first, and all HD is square pixel. HDCAM is more than likely going to be 1920×1080.
-
Lots of excellent points made here, so not to dogpile but there is one other thing I would like to add.
The Animation codec vs. Uncompressed codec comparison is akin to JPEGs and GIFs with regards to how they behave, the Animation codec being like GIFs in this example. GIF files are great with solid blocks of color, aliased text and other solid chunks, and the files can be tiny. If you encode a photo in GIF format, you may be able to get it to look okay but the file will be huge because its compression is meant to deal with solids, not gradients. JPEGs on the other side are excellent with photos, but when they are fed solid blocks they tend to have artifacts within those solid expanses.
Part two of this is that you can create a 5,000×5,000 pixel, hour long Animation codec movie that, if it only contains a solid white box, will be a tiny file. That’s because it’s compression scheme can store solid areas very well. But encode a 100×100 pixel video in animation codec and the file will be quite large.
Where the Animation codec stores solid color chunks as one large entity to save file space, Uncompressed files treat every pixel in your image as unique and therefore even large white solids will result in huge files.
So all of that said, the Animation codec is NOT the best choice for video footage. It works well for 2d animations and graphics, but only because those tend to have more solid areas without gradient information. Because the files will be huge anyway if used with video, you would be MUCH better off rendering that video footage in 8-bit or 10-bit uncompressed, which ensures no information will be thrown away or mishandled by the codec. If you are doing any effects work or greenscreen extraction on the video, I would definitely stick to 10-bit uncompressed.