1 person found this helpful
Check the position of the layers. If they're off even by 1 pixel, they can cause grief for what looks to be static bitmap captures.
Not to mention that H.264 is hardly the world's best capture codec. Lots of visual information lost in the encoding process. If you have a different option for capture, you may want to investigate it.
Hello Dave! I read a lot of your answers before asking my question, I appreciate you offering your knowledge base around. My footage in this case is 1 layer mp4 video - the cursor on top is a bitmap that moves around to direct the tutorial video I'm working on. Are you suggesting that if I had layers of <100% opacity then I'd have to be sure they were lining up? Currently the only thing I'm doing in the nested compositions is splitting out the audio and time remapping to extend certain frames. No extra layers are being brought in.
Unfortunately I have to export to h264 in the end, due to company guidelines. I don't necessarily have to record my source video in h.264, BUT the original video looks perfectly acceptable before all the nesting. I have no idea why nesting the comps should result in losing res, but it certainly looks that way to me and i don't have any other guesses at this point.
It looks like you may be using a screen capture application. True? False?
The more info we have, the better the chance of solving the problem:
True. Screen recordings taken in Camtasia - not my favorite, but its what I've got to work with. Hold on and I'll go through the whole FAQ list.
- AE version stated in post
- It was updated recently - but I don't have the admin creds I need to run updates as frequently as I like. I'm waiting on an IT ticket to run my current updates.
- Mac OSX Yosemite 10.10.1
- No error message
- This has been ongoing since I started trying to solve my script/video timing by pulling my main comp into additional comps. Created new comp from main workspace comp, copied the comp layer, turned off video for one and audio for the other so I could work with them separately. Split and time remapped video layer to synch it up the way I wanted. no other changes made I think.. gonna go double check that in a moment though.
- I've never nested comps this way before - this is the longest most complex vid I've had to make in AE, and I've never felt the necessity in the past. My videos have always looked sharp before this one though.
- Quad core 2.7 GHz Intel i5 processor, 4 gigs ram (holding out for more, I know its not ideal, but I'm sure its unrelated to this issue...right?) AMD Radeon HD 6770M 512 MB - no external drives or extra monitors
- CUDA 6.5 drivers
- no third party hardware - mac out of the box
- Source footage as stated - h264 mpeg-4 recorded in Camtasia2
- Quicktime 10.4 - im not recording with it because it doesn't have cursor removing capabilities
- Audio recorded with audacity, not much else going on besides stock programs & OpenOffice.
- no third party effects
- I am using open GL features .. I think
- Problem in both ram preview and final rendered output
- Render Multiple Frames Simultaneously is off
- I don't think I'm using the ray traced 3d renderer... I only have 2 layers in the whole project that have 3d turned on for a simple camera zoom in/out & advanced comp settings are on classic 3d.
- Problem is not with output
Heres screenshots of my main compositions:
And here's an album with all the my composition settings - the first  being representative of all my first level compositions with the video files in them.
1 person found this helpful
So I took a looked at your composition structure and your original screenshots and I have a few suggestions for your workflow and a few unavoidable facts about turning screen captures into videos.
First, the fonts on most computer screens and the lines in most web designs and app UI's are way too thin to allow any scaling of screen capture movies without terrible degrading. It looks like the fonts on your captured page are only one pixel wide so you cannot scale, especially scale down the footage without seriously degrading the quality. The other problem is that computer displays are seldom standard video resolutions. I have a separate monitor for doing screen captures. It's 1920 X 1080. You should also have the refresh rate set to 60 Hertz (NTSC) or 50 Hertz (PAL) depending on your location. I also turn on accessibility options and enlarge the fonts or scale up the browser window by pressing Ctrl and + on the mac to make the fonts and lines on the browser bigger.
On a Mac I suggest that you take a look at ScreenFlow instead of Camtasia. If you can improve your screen capture and do your initial editing of the screen capture in Screenflow or Camtasia, and render a suitable DI you'll have much better luck in post. Adding overlays, click callouts and other elements directly in your screen capture program is a lot faster than doing it in AE.
Last point. Looking at your comps it looks like you are just cutting up the nested comps. A lot of them are cuts to the same comp. I see the comps but I don't see position or scale values. I'm not sure what is going on in any of those nested comps and I'm not sure why they are not footage layers. This is what I'm getting at, you need to make sure that the screen capture footage and the nested comp is NOT scaled or repositioned. If you must reposition and your layer and both are an even number of pixels high and wide, then make sure that the position values are whole numbers. To make things easier the size of the comp should match the size of the video. Your selected footage size is 1424 X 1260 is not a standard size for that format. H.264/MPEG-4 AVC - Wikipedia It may work, but you would be better off using a standard sized monitor to do your screen captures. Whatever you do, do not scale up the screen capture to fill a standard HD composition. If the cut up footage needs to be a comp for some reason then collapse transformations should be on for all nested comps. Again, it would be very helpful to know what you are doing in your nested comps.
Last point, and I"m not sure whats going on in your composition so take this as a suggestion rather than a criticism, but it looks like your project is a lot of cuts made to the screen capture video and a bunch of audio effects like mouse clicks (which you should be able to capture), and an animated cursor. If that's what you are doing I would never approach this project entirely in After Effects. I'd put the screen capture in Premiere Pro, cut it, add my sound, animate my cursor if possible, and be done with the project. I can cut video in Premiere Pro about 100 times faster than I can cut in After Effects and if things need to be changed per client request, it's way easier. If there is something going on with the cursor that you can't do in Premiere Pro, then render your edit in premiere pro or import the sequence and add the cursor in AE.
To answer why your footage looks worse and worse I can only guess that you are shifting position or that the footage or the comp is scaled. If it were me I'd invest in ScreenFlow, do 90% of my cutting and compositing in Screenflow, then polish up the edit and add the other elements in Premiere Pro, if complex composites are required, do those in AE and bring them into your NLE to edit.
Thanks for the suggestions! I have actually looked at Screenflow and Screenflick as alternatives to Camtasia, but Camtasia just happens to be what the company already has licenses for. It's features for overlays and callouts are very very weak, but unless I can prove that's causing serious problems, I can't justify the expenditure.
The fonts/screen resolution thing is an interesting point - however in my sample renders I'm not noticing any real degradation at the lower compositions, even if the resolution goes up or down a little. The real problem only happens between comps.
Screen capture video is not scaled up at any point here either - my target size for this video is much smaller than my monitor resolution. All of my footage is at 100% in AE - except for 1 bit of footage actually that I accidentally exported larger and have at 65% here. That one is fuzzier than the rest, and now I understand why! So thank you for that!
Regarding my ... weird.. use of comps - I'm using them to time remap my previous comps to get the audio to synch the way I need. I got a request after the first version to "just slow it down a little," which is a very hard thing to do. By nesting the composition I was able to split out the video and audio and just add little breaks between sentences etc, and time remap the frames in between action to fill the gaps. That was much MUCH easier than rebuilding the whole thing in my main workspace comp, but now I'm having this problem so I'm not sure it was worth it
I could probably do this whole thing pretty easily in Premiere Pro, to be honest I just wanted an opportunity to learn more about using AE so I figured I may as well do a project in it. Not strictly necessary, but it's been a good learning opportunity.
I am shifting the position of footage/comps regularly throughout the project, however when I render the bottom level compositions with the original footage in them - changing position all over the place - there's no visible degradation. The only setting/ effect changed between the comps screen shotted in the post above is layer splitting and time remapping.
There's another reason to use ScreenFlow... you can retime in the app. You don't need to pre-compose to time remap footage. Collapse transformations would help but your biggest problem is the position at a half pixel. You've just killed the resolution of thin lines and hard edges. Time remapping rendered footage, especially if that footage is moving in the frame, is also detrimental to the resolution because frames will be blended. Make sure that you try different frame blending modes.
If I had your original footage I would start in Premiere Pro, Create a sequence named Timed to Audio or something like that. Add my footage and add my audio track, not worry about repositioning anything, put a cut every place I needed to retime something, use the rate change tool in PPro to retime the segments to match the audio. When the Timed to Audio sequence is done I'd probably start a new sequence in PPro and drag the Timed to Audio sequence into that comp, add the rest of my sound effects and animate the position of the sequence to emphasize any points that need emphasis. I would either then import this sequence to AE to add effects I couldn't do in PPro or I would add an image file for the cursor and animate the position as needed in PPro.
The degradation when you move or scale the layers is inevitable if you do not precise line up your footage with the pixel grid. Scaling needs to give you an eact multiple of your pixels, i.e. 200% so 1 pixel becomes 4 to maintain good edges on single pixel wide lines if they line up with the pixel grid perfectly. Otherwise, any scaling will degrade the image.
If I were starting from scratch I'd use an HD sized monitor on my system, Enable accessibility options to make the fonts and lines in the app I'm working on bigger. capture in ScreenFlow, record the audio track, import that to screen flow, add my annotations, mouse clicks, and other call outs in ScreenFlow, Retime the screen capture in ScreenFlow, to all my repositioning and scaling in ScreenFlow, then export a DI for polish in AE or PPro or just render the final product using the AME. You would be done in hours instead of days.
"your biggest problem is the position at a half pixel" Wow that is definitely, absolutely it. I never would have guessed that. Thank you! Also: thank you Dave for clearly giving me the right solution immediately even though I didn't understand it at the time.
You have to realize the pixels must line up with the grid or they will get reinterpreted. Single pixel wide lines especially fine text is nearly impossible to use in video. Solid black (or any fully saturated color) against white and very fine lines is also extremely difficult to compress well, especially if it is moving.
These are the limitations of working with pixels. The only way to solve the problem is to make your lines thick enough that you can hide the defects in the anti-aliasing. That is why I suggested that you enable some kind of accessibility option to fatten up the type on your page before you even start screen capture.
I hadn't even considered before (although it seems obvious now) that my position values corresponded directly to pixels. It's too bad we can't work exclusively in vectors! I'll definitely be re-reading your suggestions again before I start in recording my next project. Thank you very much for spending your time on this.