4.2.2. vs 5.02 dramatically increased file size with the same settings?

It happened that I processed the same file with exactly the same settings for enhancement and export with TVAI 4…2.2. and 5.02. The file size of the file exported with 4.2.2 is 3.5GB (17Mbps), exported with 5.02 it’s 5.7 GB (28Mbps).

Has anybody encountered something similar, and if so, any idea what this could be caused by?

1 Like

If you increase the bitrate, the file size is going to get bigger. Set the quality level or manual bitrate in 5.02 to the same as it was in 4.2.2 and see what you get.

Thank you. I used the dynamic bit rate settings shown below (my standard), because I’m unable to estimate reasonable bit rates for TVAI export with the used GPU based long GOP encoders. Nevertheless, I still have 4.2.2.1b installed side by side with 5.0.2 and will make some tests with constant bitrates and be back with information on that.

image

Is there anybody wo has with 4.2.2 installed side by side with 5.02?

FWIW I made some further tests as suggested. the file names include the version and the export settings. Highlighted my standard settings:

image

I would think, it isn’t surprising that when increasing the bitrate, the file size is going to get bigger. What is surprising, a version change doubles the file size with exactly the same settings.

General
Unique ID : 294204135756263585905736424563437723816 (0xDD55A7CDA3A43996D342335EDEB7E0A8)
Complete name : E:\Forum\HowToSlate2Edit 1080p24_Topaz_prob4auto@DynHigh_4.4.2.1b.mkv
Format : Matroska
Format version : Version 4
File size : 76.7 MiB
Duration : 51 s 93 ms
Overall bit rate mode : Variable
Overall bit rate : 12.6 Mb/s
Writing application : Lavf60.20.100
Writing library : Lavf60.20.100
ErrorDetectionType : Per level 1
VIDEOAI : Enhanced using prob-4; mode: auto; revert compression at 0; recover details at 0; sharpen at 0; reduce noise at 0; dehalo at 0; anti-alias/deblur at 0; and focus fix Off

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4
Format settings : CABAC / 4 Ref Frames
Format settings, CABAC : Yes
Format settings, Reference : 4 frames
Codec ID : V_MPEG4/ISO/AVC
Duration : 51 s 84 ms
Bit rate mode : Variable
Maximum bit rate : 16.0 Mb/s
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 24.000 FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Writing library : Lavc60.37.100 h264_nvenc
Language : English
Default : No
Forced : No
VENDOR_ID : [0][0][0][0]

Audio
ID : 2
Format : AAC LC
Format/Info : Advanced Audio Codec Low Complexity
Codec ID : A_AAC-2
Duration : 51 s 93 ms
Channel(s) : 2 channels
Channel layout : L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 SPF)
Compression mode : Lossy
Writing library : Lavc60.37.100 aac
Language : English
Default : Yes
Forced : No
VENDOR_ID : [0][0][0][0]

General
Unique ID : 17482209717917216181585549388266205786 (0xD26F37C888F418BBA7084F3E3C1F25A)
Complete name : E:\Forum\HowToSlate2Edit 1080p24_Topaz_prob4auto@DynHigh_5.0.2.mkv
Format : Matroska
Format version : Version 4
File size : 156 MiB
Duration : 51 s 104 ms
Overall bit rate mode : Variable
Overall bit rate : 25.6 Mb/s
Writing application : Lavf60.20.100
Writing library : Lavf60.20.100
ErrorDetectionType : Per level 1
VIDEOAI : Enhanced using prob-4; mode: auto; revert compression at 0; recover details at 0; sharpen at 0; reduce noise at 0; dehalo at 0; anti-alias/deblur at 0; and focus fix Off

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4
Format settings : CABAC / 3 Ref Frames
Format settings, CABAC : Yes
Format settings, Reference : 3 frames
Format settings, GOP : M=1, N=30
Codec ID : V_MPEG4/ISO/AVC
Duration : 51 s 83 ms
Bit rate mode : Variable
Maximum bit rate : 16.0 Mb/s
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 24.000 FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Default : No
Forced : No

Audio
ID : 2
Format : AAC LC
Format/Info : Advanced Audio Codec Low Complexity
Codec ID : A_AAC-2
Duration : 51 s 93 ms
Channel(s) : 2 channels
Channel layout : L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 SPF)
Compression mode : Lossy
Delay relative to video : -21 ms
Writing library : Lavc60.37.100 aac
Language : English
Default : Yes
Forced : No
VENDOR_ID : [0][0][0][0]

Do you notice any difference in the quality of the exports? Dynamic bitrate would raise the rate when needed to avoid loss, which suggests that the newer release is raising the rate because it thinks there’s more to avoid losing. You are getting more sections of your export in high rate than you did before.

No, I don’t see any difference in quality, just bloated files. I will post a bug report with media to reproduce to bounce the issue to where it belongs: Topazlabs quality control, if such an instance exists.

I would suggest trying some fixed rate exports and seeing how you like those results.

Have noticed the same thing, all my output files are much larger now using the same settings…

Thanks for your feedback.

Could you tell what your usual export settings are, including export resolution, and perhaps also which models you use?

I made some further tests and my first impression was wrong. It’s hard to see a difference when watching the exports in a media player one by one. That’s why I said there is no difference. I stacked the two clips in a timeline of my NLE and connected the tracks with a Blend effects set to difference, If there was no difference, the screen would be black, but it isn’t black. There’s a lot of high frequency noise, presumably introduced by 5.0.2, which would explain the much higher bitrate when encoding with rate control set to constant quantizer.

For people who need reproducible results that means that it isn’t sufficient to use the same settings and models, it needs to use exactly the same version of TVAI.

For people who use constant bitrate settings that means that they should double their target bitrate if they wish to get the same quality level.

Something like this should be documented, shouldn’t it?

I find, this isn’t good design, but now that it’s clear, I can handle it somehow.

This topic was automatically closed after 57 days. New replies are no longer allowed.