Ok I’m in the middle of a project for which I’ll publish a lot of video to dvd…
Right now I’m rendering everything to Avi RAW… but those files are huge… I am getting them up to 20Gig now… per file that is…
So I recon I’d better start using a codec… And I’m rendering everything in my video editor through a mainconcept mpg2 codec for dvd burning… but that’s not available in blender.
So right now I’m rendering to 1049 x 576 pixels & 25 frames a second in blender… what codec would you use??? Divx??? but at what settings??? I’m guessing Divx at 8mB a second would be best… am not sure though.
seriously, the way blender encodes to a codec [one frame at a time] seems to always result in visible artifacts, you should have a bigger hard drive if you want to render something that long
I am using 1049 x 576 because that is actually what 16 x 9 widescreen pal is when it’s uncompressed… it’s 7something by 576 when rendered, but in raw uncompressed form it’s not. ( I know calculous wise it doesn’t seem to work out, but according to my video editing software this is it)
Yes solmax… dvd is mpg 2 but you can’t render to mpg2 in blender which is exactly what I said above.
Sequenced??? are you guys ****** nuts??? Do you know how many images that would be… and btw… how would that benefit me in filesize??? 20.000 raw pictures is basicly the same as a 20.000 frame raw animation. And an animation is a lot easier to handle.
You don’t seem to get the point… I am not asking you what resolution I should use… that’s set already… I’m saying that RAW is too big… what compression should I use in stead??? and lossless is cute, but doesn’t help much file size wise… that’s why it’s lossless.
And you can advise me about buying a bigger hdd… but I have a 160Gig drive already… and that’s just not the issue. or the solution… I can fill 500Gigs of video but then my puter won’t be able to edit it anymore… That’s the real issue… I have to keep the file size down a bit so my editor can still handle it.
hm, maybe i’m not up2date, but what i know abt. widescreen is different:
widescreen (16:9) has exactly the same resolution as PAL, it just has black bars on top and below the picture. so actually, the resolution of the resulting picture is less then PAL. 16:9 tv-sets recognize 16:9 and more or less zoom in to get the full-screen-picture.
so i’d render at standard PAL resolution, which will look sqashed on a PC monitor due to the fact that it uses square pixels. on a TV monitor it will look OK again.
the codec you use - well this is up to the software you use for editing. i had very good experience with the DV-format, which blender can write via quicktime. standard DV consumes apx. 13GB/hour, which is not a big deal. Premiere can handle it very fast, much faster then common AVI codecs.
so if you ask me - i’d go for PAL DV (you could also try DVCPro, which has a higher data rate and better quality).
Yep, you can render straight to 720x576 PAL anamorphic (widescreen) by setting the AspX and AspY in Blender. No need to render out to 1049 and resize afterwards (although if you look at the anamorphic frames in an editor that doesn’t compensate for aspect ratio, it’ll look very strange indeed).
Otherwise, as arangel suggested, try the Quicktime Animation codec.
This entire widescreen thing is quite a mess… especially since once again the US and the rest of the world have different pixel aspect ratios.
All PAL (European) DVD’s use 720 x 576 pixels and all NTSC (North American) DVD’s use 720 x 480 pixels.
And for some reason (whether it’s in blender or my video editor) using the aspect ratios in blender never gets me the right result. When I render at the blender standard 16/9 PAL it doesn’t resize to wide screen at all when I put it into my video editor…
So basicly I’m stuck using square pixel rendering… and for some other unknown reason… if I want to get a screen full without stretching what I rendered in blender (deforming it)… the only size that works well at high quality is 1049/576… which should be utterly wrong, but it works. That’s the thing really… I don’t care if it’s resized… as long as it isn’t stretched. And a minor point… with blender only displaying your render results square pixel wise, it’s just not practical to render using that anyway, cause you can’t bloody see what you’re doing
Usually I’d just pick an aspect ratio that fits what I like, but I’m animating some stuff to be projected in a theatre setting and need to fully fill what the beamer can light… sorta
I’m looking to get the latest version of my editor (vegas video) so maybe in there it’s better.
Codec wise… well… divx seems to not want to work in blender… I think because it can’t divide the pixel size by 4. And the quicktime DV automaticly renders to 720x576 and messes up my aspect ratios.
Basicly unless someone has a better idea I’ll render using the quicktime animation codec, which by the way (thanks for pointing me in that direction solmax) seems to really have nice quality and makes my files less than a third of the raw size.
Long story… sorry… thanks for bearing with me ;)[/quote]
It’s a common misconception that there are more pixels in a recorded 16:9 SD image than a 4:3 one because in reality all PAL D:1 images are actually 720x576 pixels, it’s just that a true 16:9 image is anamorphically squeezed into 720x576 pixels…
720:576 is not even 4:3 because D:1 images incorporate non-square pixels to achieve the correct aspect ratio, i.e. each pixel is 1.067 times wider than it is tall, so when you multiply 720x1.067 you get 768pixels, 4/3 of the 576 frame height.
see… even if you render at 720/576 square pixel… it still isn’t quite right… since they still use an aspect ratio of 1.067
I’d seen that ratio in my editor and didn’t get it myself…
What that means though is that in relation to normal 4:3 (which isn’t what they really use)… 16/9 or 1.777 isn’t what they really use either. If I do pal widescreen in my video editor it gives me 1.4568 aspect ratio… oddness.
And that should be 839.11 to 576 pixels when square… so how come 1049/576 works?
PNG is a lossless compressed file format. So the file sequence will not be as big as raw. Instead of 1MB per frame, depending on the contents of the frame, you might get near 300k per frame.
Ok that’s nice… but… How do you work with that practically… Imagine doing a 20.000 frame long edit made up out of lots of little sequences between 50 and 500 frames long… (that’s quite normal)
Your HDD would be an incredible mess filled with images, and every time you need to add something you have to import it sepparately. It may be worth it for a short anim but as soon as you get to something like 10 mins long and multiple takes… I don’t think it’s a good choice.
For quality, use PNG or AVI raw. If you compress to a format, then convert to MPEG-2 for the DVD, you are introducing artifacts.
XVid is a format that is getting better results than DivX (the pay version) in recent tests. Quicktime is ok, but tends to be a lot slower. There may be some differences depending on the specific scenes/sources you are using.
I’d probably archive the raw renders to DVD. Use your vide editor to edit individual scenes using your hard drive and save as MPEG-2. Finally, combine the scenes in the video editor for the final cut.
I don’t think there is an acceptable shortcut if you want quality results.
20.000 frames are 13 minutes… not sooo huge. i believe its not the “logistics” behind the huge amount of files that makes things difficult but the editing software and it’s working speed. for fluent work you need a format that can be read in real time from the harddisks. also consider the final quality of your work - mpeg2 isn’t exactly high-end. so you don’t need high-res raw files, you could easyli live with something like - yes i mentioned that already - DV.
otherwise you always must wait for a preview generation - and in most programs this is what fits in your free ram.
so avoid slowing down the editing process at least as long as you just cut the material. adding effects is a different issue, and at least here you’ll end up with waiting times fro preveiws. this again is software dependent…
if you still want to use frame sequences i’d place every shot in it’s own directory. i would even organize my blend file according to the various scenes - with different scenes. each scene can have it’s own output-dir which makes it easy once you’ve set up things properly.
Solmax… agreed… and yes 14 mins isn’t much… I’m just mentioning it as a forinstance, since I did an anim like that yesterday… I’ll end up with over an hour of video on dvd… so lots of files. I’ve done this sort of thing before, but never to publish to dvd or a project of this length.
I just did my first big render in quicktime and it won’t work… too dodgy… in playback it skips frames all the time. I tried DV but am not happy with that either.
I guess I’ll have to stick with AVI Raw for my initial edits… I can render those to mpg2 and stick them together for the final product.
Oh and the effect preview thing is horrid on most apps… the fact that you have to apply an effect, then pre render, then you can see what you’ve done… That’s why I use Vegas video… it has “in line” FX… the preview in it is simply magnificent.
how long was the animation? and, do you have the latest quicktime? i’m just wondering, because i didn’t have problems, even wit longer clips. (and my current machine is… well, quite “outdated”)
I’m doing a similar project (target 4 minutes) and also facing some of the same issues, although I freely admit that this is my first project of this nature and so I am reading this post quite avidly.
I have organized each shot into its own directory. Within that directory there is an avi subdirectory, while the output of each composite-layer has its own output-layername subdirectory. Within each output-directory there are hundreds of .TGA files.
I use separate files, and this lossless format, partly so that if a render crashes, I can restart it at approximately the frame-number where it died. It’s only happened a couple of times. The final output for each scene is, for convenience, an AVI-Raw file.
So in building a scene: - First, render each layer, which creates the .TGA files in the appropriate directory. - Then, run the scene-builder which composites the various files into an AVI. - The final assembly of the film is done with a separate blend-file which strings all of the previous AVIs together.
Believe it or not, programmer that I am, I’ve found a way to automate the process with make.
Personally workflow is important for me. So I have to keep everything as simple as possible. The least ammount of files works the best. I’m close to 100 avi raw bits already… so doing layers and sepparate images would really bog down everything. I’m having enough trouble sorting through my files already hehe.
I’m glad I’m not editing the video in blender btw… it’s just not that practical is it?
that’s a lot of work man.