0
$\begingroup$

I’m creating 15 second animations with approx 500 frames (1280x720) at 24FPS resulting in 20Mb file size. Similar videos online are well under 5Mb even under 2Mb. I’m currently video editing using the default blender settings for 720p MPG4 H264 with no audio. I tried various video output settings to reduce file size but there is a noticeable drop in quality.

My initial PNG renders are at blender’s default settings but with 100% compressed RGB (no alpha) with each image about 1.2Mb. Would compressing these images further in blender or using other programs etc reduce the video output of these images? Or is there no correlation between the PNGs and MPG4 file size?

Either way, what are the ideal PNG and video output settings within blender to reduce the final video file size whilst maintaining the same quality? Or should I be compressing the videos further using an external program? These are the default video settings I am using

Improve this question
$\endgroup$

2 Answers 2

Reset to default
1
$\begingroup$

Yes...and no.

First of all, I would 100% recommend that you render at the same resolution that you plan to make your video unless you're planning on using AI Upscaling. Doing otherwise would result in the same effect that zooming really closely into an image produces - you start seeing the individual pixels.

Video files will always be smaller than the summation of the frame outputs. That's because videos don't actually store every single frame as an image, they instead use algorithms to store some frames and record what changes between different frames. Rather than trying to explain this myself here's a good short video explanation for this by Tom Scott on Youtube.

From my experience (and it should be noted that I've done some video editing using Adobe Premiere in the past, but nothing professionally), you shouldn't need to further compress the video - MPG4 will already compress the output to be much much smaller.

Improve this answer
$\endgroup$
9
  • $\begingroup$ thanks for your comments. I am using the final resolution of 720p throughout the whole process to avoid issues later on as you mentioned. Also using MPG4 because of the expectation of lower file size and widespread accessibility on more devices. So somewhere from initial render to final video output I’m not achieving the smaller files size that other similar videos have. $\endgroup$
    – Holden Chev
    Commented Mar 2, 2020 at 3:01
  • $\begingroup$ What file sizes are you dealing with, and for what length videos? $\endgroup$
    – stphnl329
    Commented Mar 2, 2020 at 4:30
  • $\begingroup$ File sizes and settings in original post $\endgroup$
    – Holden Chev
    Commented Mar 2, 2020 at 4:52
  • $\begingroup$ Huh, yeah those file sizes are rather large. You said you're using the default export settings? Mind adding a screenshot of them? But here's a suggestion that might help - the Encoding Speed setting can drastically affect output size. Setting it to Realtime will result in significantly larger files, and Slowest will result in the smallest file. It might seem a bit counter-intuitive, but I reckon that it's describing how much time Blender is spending to compress the images without losing quality. By spending more time, it can pack all the data more efficiently. $\endgroup$
    – stphnl329
    Commented Mar 2, 2020 at 5:28
  • $\begingroup$ Haha I tried the very slow encoding speed which was just as quick as the default medium speed and reduced file size by a whopping 0.1Mb :o I’m currently rendering some PNGs so I’ll send video default settings SS later hopefully $\endgroup$
    – Holden Chev
    Commented Mar 2, 2020 at 7:12
-1
$\begingroup$

A partial answer to my own question for others to add or clarify... I undertook a small sample test to compare a few variables without resulting in a noticeable reduction in final video quality of 65 frames at 720p.

Original PNGs 32 samples, 720p, 100% compression Video output settings default video file size 2.68Mb Comparisons (all settings as above unless stated) a) @128 samples 0% compression = 1.92Mb b) @32 samples, 1080p, 0% compression = 1.97Mb c) @128 samples, 1080p, 0% compression = 1.82Mb That gives about 30% file size reduction by increasing the source PNGs sampling/resolution and outputting to 720p video.

My theory here is that lower quality rendered stills will have minor imperfections or variations (eg: slightly mis-shaded pixels) not obvious to naked eye but requires a higher bitrate to process these variations than higher quality stills with fewer variations between frames.

Additionally, having higher quality original PNGs allowed a reduction in video processing settings without significant loss in video output quality which wasn’t possible with original lower quality PNGs. For example I reduced video output quality to “low” instead of the default “medium”, increased keyframe intervals above the number of actual frames and used the slowest encoding speed.

Final video file size = 0.99Mb. An amazing 37% of the original file size.

The biggest downside to achieving this unfortunately was initial rendering of each frame rose from slightly over 1 minute for lowest quality PNGs to just under 10 minutes per frame. As with many things in this field there is often a trade off somewhere depending on where your priorities lie.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .