Opened 5 years ago
Closed 5 years ago
#7483 closed defect (invalid)
CRF value not working when encoding with libaom in AV1
| Reported by: | Ewout | Owned by: | |
|---|---|---|---|
| Priority: | normal | Component: | undetermined |
| Version: | git-master | Keywords: | libaom |
| Cc: | Blocked By: | ||
| Blocking: | Reproduced by developer: | no | |
| Analyzed by developer: | no |
Description
Summary of the bug: CRF value not working when encoding with libaom in AV1
How to reproduce:
% ffmpeg -i input libaom-av1 -crf 15 -strict experimental output % ffmpeg -i input libaom-av1 -crf 20 -strict experimental output % ffmpeg -i input libaom-av1 -crf 25 -strict experimental output % ffmpeg -i input libaom-av1 -crf 30 -strict experimental output
ffmpeg version: ffmpeg-20181007-0a41a8b-win64-static
built on: Windows 10 64-bit
No matter what CRF value is used, the encoded files will have the same bitrate.
Here is the file I used to test (VP9) and the encoded results (AV1) with identical bitrates:
ffmpeg -i C:\Falcon9Timelapse360p.mp4 -c:v libaom-av1 -crf 15 -strict experimental av1_test_Falcon9Timelapse360pcrf15.mkv ffmpeg -i C:\Falcon9Timelapse360p.mp4 -c:v libaom-av1 -crf 20 -strict experimental av1_test_Falcon9Timelapse360pcrf20.mkv ffmpeg -i C:\Falcon9Timelapse360p.mp4 -c:v libaom-av1 -crf 25 -strict experimental av1_test_Falcon9Timelapse360pcrf25.mkv ffmpeg -i C:\Falcon9Timelapse360p.mp4 -c:v libaom-av1 -crf 30 -strict experimental av1_test_Falcon9Timelapse360pcrf30.mkv
Attachments (7)
Change History (11)
by , 5 years ago
| Attachment: | Falcon9Timelapse360p.mp4 added |
|---|
comment:1 by , 5 years ago
| Keywords: | libaom added |
|---|
comment:2 by , 5 years ago
Maybe I jumped too quick to my conclusion, there is a little bit difference in bitrate between the different CRF values. I expected the difference to be way bigger since the range should be exponential. There should be roughly a factor two in file size difference between each file.
I'm now rendering additional video's with CRF values of 10, 40 and 50 to better understand the curve.
comment:3 by , 5 years ago
It looks like the encoder hit a ceiling around CRF 20 / 393 KB, where a increase in quality could not be realized even with a higher bitrate, because all the detail of the source video was already maintained. Next time I should use a higher quality source video.
This issue can be closed.
comment:4 by , 5 years ago
| Resolution: | → invalid |
|---|---|
| Status: | new → closed |



Source Video (VP9)