32 bit vs 64 bit render times big diff?

80688068 Posts: 166
edited December 1969 in Carrara Discussion

Hi
As I've posted before, I like to use Quicktime Animation codec.
Can't in Carrara 8 64 bit, so I guess I need to render in 32 bit version.
Any data in the difference in render times, that might sway me to use a different codec, and stick with 64 bit?
Thanks
8068

Comments

  • ManStanManStan Posts: 0
    edited December 1969

    Yes, significant especially if you have a multi core CPU.

  • 3DAGE3DAGE Posts: 2,003
    edited December 1969

    Don't use a compression codec,.

    Use sequenced imaged,.

    The time that it takes to render each frame, is the same, whether you then save it to a file, or add compression and encode it as a movie format,.
    Saving or encoding only happen after the image is rendered fully. ..so it doesn't effect the rendering time.

    The only difference is that the "Movie" (depending on the codec) may have compression,. the image, (depending on the image format) won't.

    There should be some difference in the render speed of 64bit / 32bit. but (depending on your system, and the scene) it may not be a significant difference.


    You should try to keep the Highest image quality you can,.. until you're ready to encode into a final movie format.

    :)

  • evilproducerevilproducer Posts: 7,577
    edited December 1969

    3DAGE said:
    Don't use a compression codec,.


    The only difference is that the "Movie" (depending on the codec) may have compression,. the image, (depending on the image format) won't.

    Hey Andy, this bit doesn't make sense. As far as I can tell, most of Carrara's still image formats are compressed. Not all are lossy, but neither are all the video codecs. You have just as much a chance of selecting a lossy image format to render to as you do a lossy video codec. if you render to an image sequence, you will still have compression. As I'm sure you're aware, there are codecs that are considered lossy and lossless. In other words, for a lossy codec to achieve a smaller file size or lower data rate, the compressor throws out information. Some do it better than others. Images also use compression. A gif is a compression format, as is jpeg. They're both lossy, but jpeg is mnore modern and much better than gif, bit for bit (in my opinion). I believe another lossy file format is png, which is more modern than jpeg, and supports alphas, which jpeg does not. Common lossless image compression would include tiffs, RAW, Photoshop and maybe Pict. (not 100% sure on that last one.)


    The advantage to rendering to an image sequence is not that the video codecs are lossy and the image formats are lossless, it's that if Carrara crashes halfway through an animation, you don't lose the work that was already done and you can continue from the point where Carrara crashed.

  • evilproducerevilproducer Posts: 7,577
    edited December 1969

    Forgot to add that if you're on Windows and don't have access to the QT codecs in 64 bit (or 32 bit) directly through Carrara, then rendering to an image sequence is the way to go, because then you can open the sequence in QT Pro and export to your desired codec. As a test, I rendered an image sequence with an alpha channel to png, opened it in QT Pro and exported it using the Animation Codec and then composited it in Final Cut Pro. The alphas were supported throughout the whole process, and I didn't see any compression artifacts, despite the trans-encoding.


    The process was a bit laborious. If I'm rendering a scene with an alpha channel, it's because I'm simplifying a complex render anyway, and I'll just render directly to a movie instead of sequenced images. If I have trouble with Carrara choking on the scene, then I'll go and render to an image sequence.

  • 3DAGE3DAGE Posts: 2,003
    edited December 1969

    HI EvilProducer :)

    You're quite right,. ...I'm saying it wrong.

    Depending on what image or movie format you save to,. you'll either loose information, ...or not.
    If you're going to make the footage into a longer movie, then you'll want to stick with lossless formats.


    Compuserve GIF format has a limit of 256 colours and alpha,. so,. the file size is reduced by data loss,.. converting millions of colours to 256.

    Portable network graphic (PNG) is great since it handles Millions of colours plus alpha, but it still uses jpeg compression.

    Jpeg (depending on the quality you save at), can also be just as good as PNG, but without the alpha.

    so,. PNG or JPEG would be ideal for a draft render. (and even a final render) depending on the settings for each format

    Photoshop files save all the image data,. plus multi-layers,. and, (as far as I'm, aware) encodes that into a format that's compressed.

    TIff is also compressed (I think using the LZW compression) but it retains all of the image information.

    so,. those would be better for a high quality final render.

    You can save a movie in an uncompressed format, but they're usually Huge files,. and that makes working with them harder.

    Sequenced images are editable individually so, there's also that advantage for post production on individual frames.
    An image format also has the advantage of allowing you to use Carrara's Multi-pass rendering to export different scene information.

    so, if you used photoshop PSD , and multi-pass rendering, then you can open that sequence of frames in After effects and work with the different layer information.

    As you say, If it all goes crunch while you're rendering,. you only loose the frame that was being rendered when it crashed.
    and you can pick up from there, instead of rendering the whole movie again.


    If you're gong to work with your footage in a video editor, then by using a compressed "lossy" format to start with, it will become worse when it's exported as a final movie with even more compression ( double loss).

    does that make sense :)

  • evilproducerevilproducer Posts: 7,577
    edited December 1969

    Much more accurate- And surprisingly less verbose than my initial reply ;-)

  • holly wetcircuitholly wetcircuit Posts: 0
    edited November 2012

    3DAGE said:
    As you say, If it all goes crunch while you're rendering,. you only loose the frame that was being rendered when it crashed. and you can pick up from there, instead of rendering the whole movie again.

    Certainly not quibbling about good advice! :cheese:

    :red: except this one little bit which I will clarify...The reason you have to render all over again from the start is that the video file will (probably?) be corrupted or unfinished, causing you to lose the work you've rendered...

    Carrara allows you to select a start and end frame when rendering to video formats... So there are other strategies as well, such as rendering in 5 or 10sec bursts for example. I know a lot of people recommend rendering to TIFF, but in my day-to-day practical world TIFFs are way to large to store and too slow to use for video (your NLE will likely re-render it into an intermediate format just to be able play it, adding even MORE files to your harddrives...).

    If you *are* crashing during render, you should consider the reasons for the crash.... Is the scene too large? Is there a "memory leak" where you run out of free RAM? Is there plenty of room on your HDD? (most large scenes will start to use Carrara's virtual memory, which means writing to and reading from the temp file constantly....), or overloading your scene with n-degrees of model smoothing, subdivision, or displacement.... Oversize and redundant image textures will also bog up memory and slow everything down. If you are serious about animation, cut the fat.

    I just want to say that I use Carrara constantly for very short animations ("sprite" elements for motion graphics, as opposed to rendering a huge 3D forest scene with full GI etc). Rendering directly to Animation codec is the best format for this purpose. Carrara is very stable, until it isn't.... If you are experiencing ANY regular crashes during rendering you should re-think how you are constructing your scene.

    Post edited by holly wetcircuit on
  • holly wetcircuitholly wetcircuit Posts: 0
    edited November 2012

    oops...

    Post edited by holly wetcircuit on
  • GFHELGFHEL Posts: 24
    edited December 1969

    3DAGE said:

    Portable network graphic (PNG) is great since it handles Millions of colours plus alpha, but it still uses jpeg compression.

    I think PNG uses lossless compression. At least this is what the W3C states in the specification: http://www.w3.org/TR/PNG/

  • evilproducerevilproducer Posts: 7,577
    edited December 1969

    That's cool. I thought since it had been made a web standard that it would almost have to be a lossy compression.

  • ChoholeChohole Posts: 19,378
    edited November 2012

    I didn't think PNG were compressed at all as I get such clear images from PNG

    Post edited by Chohole on
  • 3DAGE3DAGE Posts: 2,003
    edited December 1969

    HI GFHEL :)

    I think PNG uses lossless compression. At least this is what the W3C states in the specification

    quite right,. my bad,. it's not Jpeg compressed. my bad memory

  • evilproducerevilproducer Posts: 7,577
    edited November 2012

    chohole said:
    I didn't think PNG were compressed at all as I get such clear images from PNG

    You have lossy and lossless compression. Apparently pngs are lossless, but they're still a form of compression. I don't know how it all works on a code level.


    As an experiment, and to show that there is compression involved even for lossless formats I took an image that I had rendered to Photoshop. I flattened the image and saved it as Photoshop, tif, png and the lossy jpeg. Here's some screen shots of my Get Info windows showing the various file sizes:


    added the image for reference.

    Moss.jpg
    2000 x 1500 - 1M
    Picture_4.png
    264 x 163 - 19K
    Picture_3.png
    268 x 167 - 20K
    Picture_2.png
    265 x 165 - 18K
    Picture_1.png
    263 x 162 - 18K
    Post edited by evilproducer on
  • GarstorGarstor Posts: 1,411
    edited December 1969

    ManStan said:
    Yes, significant especially if you have a multi core CPU.

    The key difference between 32-bit and 64-bit is in the amount of process address space that can be accessed without any "dark magick" trickery. I did a series of posts about troubleshooting PC performance on the old forums.

    32-bit limits your process (i.e. Carrara) to 2 GB of address space. Even if you have 8 GB of RAM, Carrara won't be able to use more than 2 GB (keeping this simple by avoiding the aforementioned "dark magick").

    64-bit gives processes a whopping 8 TB of address space -- only massively parallel super-computers have memory that large. The most I have ever worked on is 512 GB of RAM (pretty sweet).

    Of course, multi-core gives your system a lot more processing power; but that power is not related to the "bit-ness" of the machine. More addressable process space and more physical RAM (i.e. 8, 16, 32 GB or whatever) means that Windows does not have to swap data between memory and the hard drive nearly as much -- that is a big performance gain no matter what the bit-ness or number of cores in your box.

  • ChoholeChohole Posts: 19,378
    edited December 1969

    chohole said:
    I didn't think PNG were compressed at all as I get such clear images from PNG

    You have lossy and lossless compression. Apparently pngs are lossless, but they're still a form of compression. I don't know how it all works on a code level.


    As an experiment, and to show that there is compression involved even for lossless formats I took an image that I had rendered to Photoshop. I flattened the image and saved it as Photoshop, tif, png and the lossy jpeg. Here's some screen shots of my Get Info windows showing the various file sizes:


    added the image for reference.



    Hmm, that's interesting, thanks. I like PNGs, work well for me, and have automatic alpha channels.

  • holly wetcircuitholly wetcircuit Posts: 0
    edited December 2012

    As an experiment, and to show that there is compression involved even for lossless formats I took an image that I had rendered to Photoshop. I flattened the image and saved it as Photoshop, tif, png and the lossy jpeg. Here's some screen shots of my Get Info windows showing the various file sizes:

    I always use PNG unless I need the Multi-Pass layers....

    Look how PNG is half the size of the PSD.... And TIFF is just horrible! Almost TWICE the file size of the PSD.

    Screen_Shot_2012-12-01_at_1.52_.42_PM_.png
    558 x 353 - 60K
    Post edited by holly wetcircuit on
Sign In or Register to comment.
Rocket Fuel