I have multiple series of images that I want to transform into video. The images and series have the following characteristics:

  1. .PNG format. Video format not important, so long as it is popularly supported.
  2. All images in a series are the same size.
  3. Different series have different image sizes.
  4. Image sizes range up to 4800*4800 pixels
  5. All images are 8bits/channel RGB
  6. In a particular series the images change very little from one "frame" to the next - typically much less than 1%.
  7. Each series consists of 100-1500 images.
  8. All images in a series have the same colour map and registration - there is no adjustment for either of these characteristics from frame to frame.

I am looking for an open source tool or cheap COTS tool to apply to these series to turn them into videos. The tool needs the following functionality:

  1. Able to take advantage of the high commonality of image from frame to frame, to maximally compress the resulting video.
  2. To take such a series and convert it into a video.
  3. Run on Windows7 on four core 64 bit processor.
  4. Accept creation date, or file name sequence as the frame order.
  5. While there is no requirement to maintain the image depth, individual frames must not be manipulated (compression, colour flattening; etc) so much that detail is lost. If the tool adopts any of these approaches, the degree to which it is done must be user controllable.
  6. Allows the frame speed to be altered within a series of images. (I might want the first second to be 13 frames; the next 64 frames; and so on).
  7. Each series is to be turned into one video - there is no requirement for any advanced video editing capability.

I have done a little preliminary investigation and have found this question, which I am investigating, but is in respect of Linux. In addition I have Irfan graphic viewing and editing software which while it looks very useful, on first use, does not seem to meet all the requirements listed above.

Any suggestions for a solution that meets my requirements?

EDIT: Progress Report I have had the following suggestions:

  1. @kusma (on SO) suggested VirtualDub
  2. @rems (below) suggested ffmpeg.

On starting to investigate these suggestions it has become apparent that my original question is much more complicated than I anticipated. In particular:

  1. it appears that most devices will have problems with image resolution of 4800*4800
  2. variable frame rate is not permitted by AV1, which mandates that the frame rate be fixed
  • Please don't crosspost questions, when the question is closed on SO, it will automatically be moved. If you want to move it "immediately", click "flag" at the bottom of your question on SO and request a migration to SU. – Pylsa Feb 15 '11 at 11:52
  • @BloodPhilia - sorry - I was not aware of this protocol. – Chris Walton Feb 15 '11 at 11:56
  • Never mind! I have already requested migration for you. – Pylsa Feb 15 '11 at 11:59
  • Maybe you have a look at sourceforge.net first. – Michael K Feb 15 '11 at 12:25
  • Your question on SO has been closed, so this question can just stay here! ;) – Pylsa Feb 15 '11 at 12:41

Try ffmpeg from http://ffmpeg.mplayerhq.hu . There are precompiled binary packages at http://packman.links2linux.de .

But you will have to compile it on Windows, see FAQ http://www.ffmpeg.org/faq.html#TOC35 .



Just do something like:

convert -delay 5 *.png movie.mpeg

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.