Video Formats For Animations - Which One?
Steve K
Posts: 3,191
I enjoy doing five minute animations in Carrara for the 48 Hour Film Contest. The contest rules specify either Quicktime or MP4, and no proprietary codecs. They also recommend that you submit a backup copy of your video (flash drive or data DVD) to ensure they can play it. In the past, I have submitted a Quicktime and an MP4, but have noticed problems with frame proportions with video clips from different programs, e.g. Carrara and VUE. I think the Quicktime rule is mainly for those using Final Cut Pro, not available for Windows. Plus Apple no longer sells Quicktime Pro, so I have it only on my laptop and thus can't see it full resolution on my main animation desktop (the non-Pro version shows only half resolution). I did find a freeware app that will play full resolution - VLC media Player - but its still a hassle.
So. This year I abandoned Quicktime and used only MP4, submitting both a flash drive and a data DVD. No problem, the local producer got it to look good on the big screen. My video editor, Magix Movie Edit Pro, will do both plus Windows Media Video (*.wmv). I actually would prefer wmv, since the files are much smaller, but the contest folks seem way less than pleased about wmv. "If you five minute video file is less than 1GB, you've probalby made a mistake" they say, mostly about live action video. My video friends thinks that's crazy, but the mov and MP4 files do come out about that size - a little less for animations since they compress a little more, but not much. I confess that is usually convert Carrara footage to wmv for dropping itno the eidtor, since it plays back with no delays, vs. huge uncompressed avi clips, which is about all I can get Carrara to do (only a couple of really old codecs besides). And yes, I know a JPG sequence is preferred, but for a two day contest that's just one more complication.
After all that long winded background ... what do you folks like to use?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
EDIT I just just discovered 3ivx does not have a 64-bit version. Hmm. Sorry.
Sequenced PNG renders. Imported into video editor tracks. Rendered as 720p in MP4 or MPG format, depending on the video editor I'm using. YouTube loves both formats.
I do image series compiled into uncompressed avi using virtualdub if using Windows live moviemaker for wmv or Handbrake for mp4
the latter for my 360 as need mp4 to inject metadata
Otherwise as a folder of images in Hitfilm but the mp4 are bigger there as I have no idea about bitrates etc and too low is jumpy there.
Sequenced PNG definitely. And also, in my opinion, you should always be rendering out in 3840 X 2160.
Hi Steve,. :)
You shouldn't see sequenced files as a complication,. see them as "frames" of footage,. which is exactly what they are,.:)
JPG (if you're including the background) or PNG if you're including any alpha)
either should be easy to work with in any video editor,.
Export as MP4 (H_264 codec) if available,. or quicktime (jpg)
my preferred settings are 1280x720 ,.
I see no need to render at 4K unless you're being paid to,.
Try Lagarith Lossless!
Okay, thanks for all the feedback. Lots to consider here. The image sequence is a good suggestion, although I've had no problems with video formats. Also, I understand the larger frame size preference in most cases, but in a two day contest, the render time would increase. In most cases, I'm getting 5 to 10 seconds per frame render time. As I mentioned, my 720x480 videos look fine on the medium size theater screen at the screenings, so I don't see a lot of benefit to larger frame sizes for this specific contest.
I always render to sequenced PNG, then compile to AVI using VirtualDub and usually then convert to MP4 as it is so much more compact.
Thanks, Phil. For the 48HFP contest, I have settled on MP4, which seems to work fine for the theater screenings - no problems with clips having different ratios, etc. For other purposes, I might still use wmv, since the file size can be pretty small, albeit with somewhat less quality. Here is a brief comparison (I don't know how reliable this website is, but it looks legit):
http://www.iorgsoft.com/compare/mp4-vs-wmv-comparison.html
I use VirtualDub to convert sequenced PNG straight to MP4.
One person's approach:
Starting with at least 24fps PNGs saved/converted to the (windows only) lossless lagarith AVI (RGBA!, not the YUV options if can avoid) using any decent editor, converter, or virtual-dub/avisynth tools to create lossless alpha-smart AVI clips for future editing and clip management. These lossless files are a bit heavier (10-15 times good quailty MP4 file sizes). This is basically 'zipping' a folder of lossless PNGs into a 'lossless video blob' (called an AVI in this case) for editing and handling.
Then from the video editor, export interim results to those same lossless formats, and export final results to MP4, generally rendered at 24 or 30fps around 12,000bps bitrate, 1080p (1920x1080) or 8000bps bitrate, 720p (1280x720). Up the bitrate if your scenes are high-motion and you see mpeg 'relics', lower if fairly static scenes or cel animation. sound at least 44.1k, or 48k if you can. These are starting points... on a big wall, a gig of video for 5 minutes might be what you have to do w/1080p/720p. (for reference, blueray MP4 is running 29000bps vs DVD's 10000bps -ish IIRC)
Others have recommended always render larger than your target output - sure, if the render-time and storage space trade-offs work for you, and if you intend to 'zoom' in with your editing (rather than using the render camera to zoom, etc.) go for it.
All lossless (or near-lossless) codecs I know of seem to run 10-15 time good quality MP4 file sizes, and for the little bit of extra sparkle, Lagarith, Huffyuv, Cineform, etc. might be overkill for most projects - but if you need that extra clarity, and you've already spent the rendering energy to get the pixels in your PNGs, you might be glad for the option.
Note that a key element of this data-flow is to keep the best possible image sequence going into your editor(s), saving as 'lossless' clips and sub-scenes until you're finally rendering for presentation/distribution with titles, transitions, zooms, color grading/adjustments, effects, blurring, etc. If you can't get the lagarith or huffyuv codecs (macs), the GoPro Studio Cineform codec is pretty good and portable to Macs/PCs, and given it's GoPro, should be useable to others in your workflow - I think it does alpha as well if you tell it too, but adds 25% for that channel as you might expect, and is also fairly hefty compared to MP4. The GoPro Studio is free at gopro.com and teh codecs come with the install (I think they have a mac version!) - Iclone has a newish internal alpha-capable format for their toolsuite compositing but it's only them for now. For maskless compositing/alpha options, the usual chromakey tools are all over the place, so you can get around alpha issues with most any tools with some tricks and planning.
Note that some of the video 'cutters' will convert, cut, and reconvert back, injecting two lossy passes into your clean newly cut video. They never seem to leave the sound alone either. urg. AVS4YOU has a $50 suite that includes a 'video remaker' tool that claims to cut on 'GOP' boundries, and saves the same frame sequence as the original - lossless - seems to work in the cases I've used it, and tries to leave the sound alone if it can... There's another lossless cutter out there that comes up early in the google searches, but I have no experience with it other than it says it's doing exactly the right thing as its biggest selling point.
Most folks don't know that their videos are almost always converted to RGB when loaded into their video-editors, so when they start out non-RGB, even the first generation edit injects 'loss', then they are usually saved back to a 'YUV' format for a lossy second hit, to be loaded again later for a third, etc., which is why I recommend the PNG sequence to lagarith RGB or cineform RGB formats (RGB is not the codec default). That way they start as RGB from the (carrara) render 'camera' into lossless PNG, into lossless RGB AVIs and load in the editor as RGB... clean. A final implication of that flow is any editor that can easily load a folder full of PNG image sequences is also lossless RGB in the right way! But those folders are a bit more work to toss around (zip is your friend). If the codecs are installed, most editors can export these edited clips back to the various AVI RGB formats (Corel loses the alph on export). Also few folks know that JPG sequences are functionally 'YUV', PNGs are usually generated lossless = RGB. JPGs are fine for alpha masks, but lossy from the start if you're thinking about rendering to jpeg sequences. Smaller than PNG, but not that much smaller. Rendering jpeg mask layer image sequences ought to be fine though. Animation (cel style) might be OK with jpeg - discussions abound on that.
Lastly, I shy from non-open corporate-driven standards like WMV and QuickTime even if they are admittedly solid performers. I will render finals to those if I know the target audience is tuned for those formats, but otherwise tend to favor the open standards organization and de-facto open-ish formats - MP4/x264 this week. With the 4K,HVEC,x265, who knows where it'll go, but I know all of my currently rendered source sequences will be as good as the original renders were, regardless the output targets tomorrow. I hope it was worth the trouble!
As a parallel workflow, with sound, I always convert everything I get to PCM/wave at the original bitrate for any processing/editing/cleaning, saving intermediate clips to the exact same settings as the source. I even extract sound from camera-based video, clean/tweak, and re-attach to the video sometimes. Ironically, for the final render/edits I drop stuff in the video editor and simply let it upconvert or down-convert to the project's render output. There may times where manual conversion in a sound editor might produce a better result (better algorithms?), but my key starting sounds/music are usually already at the project target rates (44.1k+) and the occasional sound effect may need conversion to match with no issues so far.
So... if you're OCD about video/sound loss, take this stuff to heart. If not, render to PNGs, load them into your editor as a sequence, and try to edit everything right from that master project, and most of what I describe here won't be missed. When you start interacting with others and mixing projects and stock video, and... this stuff is probably handy to know.
FWIW, I use avisynth, virtualdub, MAGIX MEP and ProX, and Corel VSPro, as well as codecs/converters from AVS4YOU, infognition, utils: ffpmeg, x264, ImageMagik. I have hitfilm, and wish I'd started with Sony Vegas based on reputation, but all of my tools have good capabilities and each has sweet spots.
Sound: Audacity, AVS4YOU tools (converters/cutters), screaming bee morphvox, audio4fun VCS diamond, Magix MusicMaker and Samplitude
Even if all of this finery is going to be uploaded to youtube for 10 million views... at 480p on an old iphone in a noisy subway station... all good.
... and to the OP, I think WMV is fine as a final output. Not so keen as an intermediate storage format, but likely as good as any non-lossless format if you kick up the quality settings.
Hope this is useful to somebody or two.
cheers,
--ms
Mindsong, I did not see you mention one pass or two pass encoders in your post ... do you use two pass encoders?
@cordoni, as you know, that would be for the MP4 final result output stage (x264 setting), and buys you a better encoding at the expense of the two-passes -> analysis + encoding => slightly added complexity and some additional encoding time.
If you want a better result, I'd say sure..., fire it off and get a cup of coffee and do something else for a bit - or do it before bed... Maybe try it on a small representative clip to see if you can 'see' any difference in your specific content. If I wanted to do a final for a 'real' DVD/BR disk product, I think I would probably enable that just to be sure. The good news is that the resulting MP4 is simply better quality, with no computational cost to the viewing decoder/device.
One point of note, is with the higher resolutions/bit-rates, regardless of format (MP4/WMV/etc.), you have both more data (megabytes/gigabytes) to move, and computational overhead *per frame* during decoding/presentation. If you find the device that you're presenting on (PC, phone, Roku box, TV or DVD/BR player w/ USB or network stream) is 'stuttering' while playing, you may have to drop the bitrate. It may be computation or network-stream, but lots more info means lots more to transfer and process... older devices especially.
To those who know this stuff already, I hope my description/tone above were not too sloppy/condescending, as I think the big ideas are valid, and knowing roughly why/what you're trying to do is better than any specific encoder or tool that might soon be obsoleted by a better one.
It seems wrong to render out brilliance and surrender any of that to unnecessary conversions or loss.
I've occasionally taken crystal clear video and added softening and grain effects in the edit stage to 'date' it and make it less 'perfect', as the crispness seemed to distract from the experience. Ironic to work that hard for 'perfect', then intentionally 'smudge the lens'..., but having the choice and control to do so by *choice* was great.
hth,
--ms
If your inputs are 'small' you've probably surrendered some of the quality from the start (my OCD screams nooooo! :) to make your editing process useable - understood. So, me being a user and liker of Movie Edit Pro (MEP), I recommend that you be sure you enable the 'proxy' editing feature, and make sure it's enabled in the little blue lightning bolt under the viewing monitor window.
This proxy feature is available in most medium-tier video editors now, and is designed to resolve that specific problem by rendering a mini/fast version of the movie for the editing stage, and then rendering the real thing full-resolution.
There are a few other proxy resolution adjustments and effects disabling options in MEP (and others) to help speed things up even more, but the proxy feature is the big win.
Again, most video editors have this feature now, as having enough system memory or memory bandwidth available for managing the uncompressed pixel data present in the latest 4K videos on our motherboards isn't something most of us can afford. Try saving 10 seconds of *any* (even all black) 1080p video to the 'uncompressed video' option of the AVI format, and you'll appreciate the magic of modern compression. Google just came out with a new compression spec called V180 for their youtube VR stuff. When you consider the implications of 180 and 360 degree video at the resolutions we're now expecting... heh, it's going to be interesting . Right now, most of our video is displaying a static 70-90 degrees Left/Right/Up/Down rectangle - what, 15% of 360 surround? (picture a decent resolution movie projected on a full Carrara skydome or 30+ 1080p TVs surrounding you...)? wowzer...
--ms
When I see the price of a WD RED 3To (100 Euros), I don't hesitate to compil my images sequences in Uncompressed AVI (with or without alpha), a lot of giga's, but the best quality to work in AE.
Each rendered images sequence (I prefer .TGA) pass through After Effects to make a final footage.
I don't choose .PNG because it was a very compressed format for the internet, it could be better now?
My final mounting is in 720P, an excellent format for the theaters, and my original renders are in 1050x576 images sequences.
The main thing is to work on the best quality before the final compression and/or upscaling.
For the theaters, it's MPEG 720P/48Khz and I upload my short tests on YT in uncompressed AVI format.
I compose my musics and soundtracks in Cubase 5.
Each one has his preferences...
whew - The *only* reason I chose PNGs was I was led to believe that they're lossless (and have alph/gamma info support). Your note gave me a scare, so checking the libpng page (http://www.libpng.org/pub/png/pngintro.html)
I am calmed by:
"Since PNG's compression is fully lossless--and since it supports up to 48-bit truecolor or 16-bit grayscale--saving, restoring and re-saving an image will not degrade its quality,"
So if that's the case (seems to be to my OCD/experience), I'm quite happy. TGA and TIFF have good support and history, so no problem with either, other than size for me.
I wonder if the PBM format that I think came out about the same time is what you bumped into.
I like yours. Good choices and good tools. Do you think you'll kick it up to 1080p anytime soon? I'm there now, but not planning to go to 4K in the foresee-able future...
I'd also like a WD Red 3To for 100 Euros... maybe a couple of 'em!
Good to see what others are up to!
--ms
I imagine at some point (depending on the level of compression) it may take longer to load/save the compressed frame than the uncompressed version ... haven't tested this though.
ooooh, that gets the gears going - the key bottleneck today, even with SSDs, would be the rate of data from the disk. So a small well-compressed file gets read from the disk (quick?), uncompressed and moved over the memory bus to the application's memory - it's fast in the app from there on. Uncompressed data is read directly to app memory from the disk, no CPU-time, but a *lot* more data, and it's identical in the app after that... disk loading time is key, and we need to remember that while the CPU is uncompressing the data, the file loading/buffering is also happening in parallel, making the apparent load time almost moot in most cases. As compression gets more expensive, the disk loading becomes even less of an issue for the compressed model.
It's when disk storage is the same as memory that the game really changes, right? nothing has to move at all. instant-boot systems are touching on this already. price is the driver, I suppose. I bet it happens in the next twenty years. Maybe ten?
so yeah, the compression is still faster for now, but the faster the disks get...
Moving stuff across most networks will still benefit from compression for a long-time-to-come, I'd think. Even fibre to all desktops would suffer the bottlenecks between sites and cost issue?
That said, I have enough problems with backups and exterrnal disk management (or the lack thereof...) with my compressed stuff - as cheap and large as disks have gotten, multiplying data size would *kill* me, even if the computer/apps didn't care either way.
this stuff is great!
--ms
Hi MindSong!
I'm an old dynosaur of the cinema, 98% of my career was held with the silver haleed films.
I passed to DVCam because the projections in the theaters were of good quality but of a very synthetic aspect, one didn't have the warmth of the ancient footages.
Now, I bought a professional camera HD… the result is even worse in my eyes!
I cannot imagine in 4K…
To come back to our passion, the synthesis animation, I have a big experience of projections in theaters, I take part to jurys during 3 months each years and I can confirm that our animations are of highest quality that the traditional films in HD, this is why I allow upscaling my images which are better in projection.
So excellent that I had to make the aquisition of a plugin for AE which render a “Cine-like quality”!
A thing is shure: there is a “technological” quality and there is a “human” quality, my eyes prefers the second one…
Have a nice day (or night)!
That sounds like a good suggestion. Unfortunately, my version of MEP (Magix Movie Edit Pro MX Premiun Download Version, 2012) apparently does not have the "proxy" feature, it seems to be in the current "Plus" and "Premium" versions. so I'll have to see if Magix comes along with an upgrade deal ...
Thanks for the tip.
DUDU, have you tried any of the LUT for the camera on the footage the presets that have a wide range of 'films' to adjust digital camera looks to?
I think the HitFilm Express 2017 (free version) has that I know the pro one does. on proxy
Yeah... Express has Proxy ;)
So, for those of you using folders of sequenced images, what do you use to casually view your sequences with... your editors? One of my favorite reasons for rendering to avi is for the simple double-click playback. I cannot just drag a folder of image sequence into Media Player Classic, for example.
Do you have Pro? I can't wait! One day... One day....
Just checked, yes there is an upgrade version. You might note that thru the end of July you can get your version for $10 more than the standard version, and also has a few bonus plug-ins, I couldn't check to see how much the upgrade is. Thanks for mentioning it... might just grab it.
I missed this one, @DUDU, sorry!
I also hail from the era of silver coated cellulose. I miss it it some ways, but enjoy the infinite options and control of this 'new' world. (too many options!)
I'm trying to identify why something in real-life can be 'crisp' and not bother us (uncanny valley, etc.), and the same thing on a 4k Samsung screen is gorgeous, but not quite right to my eye.
I think there's a natural 'bloom' to edges (almost a fresnel effect) that our eye physics produce, where, like haze, the 'bloom' or edge-blurring varies with brightness and distance.This doesn't happen quite the same way with good cameras and renders, so we look at a hi-rez screen with all of the elements of depth (foreground, layering, haze, etc.), but it's all at a single depth from our face/eyes, and the 'bloom' factor is therefore constant for that single depth, and therefore not 'sync-d' with the depth cues we're used to. And our brains aren't sure what's wrong... That's my theory, so far...
I could see a photo-shop plugin that mixed a depth-map and bloom effect making an image more 'correct'.
That said, i want to like the hi-rez stuff, as it's quite impressive, but I can't really say that like it yet. Maybe I'll get used to it.
cheers,
--ms