PNG Compressors

[Jay123210599]

I've found these two image compressors online? Do they give lossless results?
https://saerasoft.com/caesium/
https://www.jpegmini.com/

https://www.youtube.com/watch?v=-rp64paT13M
https://www.youtube.com/watch?v=i1_AkG7QSMs&t

[poisondeathray]

I've found these two image compressors online? Do they give lossless results?
https://saerasoft.com/caesium/
https://www.jpegmini.com/

https://www.youtube.com/watch?v=-rp64paT13M
https://www.youtube.com/watch?v=i1_AkG7QSMs&t

No

[Jay123210599]

I've found these two image compressors online? Do they give lossless results?
https://saerasoft.com/caesium/
https://www.jpegmini.com/

https://www.youtube.com/watch?v=-rp64paT13M
https://www.youtube.com/watch?v=i1_AkG7QSMs&t

No

Can I see proof?

[Cornucopia]

Do a bitwise comparison (of the payload). If the resultant round-trip is identical to the source, it's lossless. That is literally the meaning of the term.


Scott

[poisondeathray]

JPEGMini is lossy

"perceptual quality" is a synonym for lossy

"without visible quality loss" is a synonym for lossy

This technology is integrated into JPEGmini Pro allowing users to reduce the file size of their photos and videos with limited quality reduction. JPEGmini technology is built around a perceptually aligned image quality measure, which reliably determines the maximum amount of compression which can be applied to each individual image or video frame without introducing visible artifacts

I'm wrong about Caesium - it offers a lossless option , but I didn't test it .

[Jay123210599]

JPEGMini is lossy

"perceptual quality" is a synonym for lossy

"without visible quality loss" is a synonym for lossy

This technology is integrated into JPEGmini Pro allowing users to reduce the file size of their photos and videos with limited quality reduction. JPEGmini technology is built around a perceptually aligned image quality measure, which reliably determines the maximum amount of compression which can be applied to each individual image or video frame without introducing visible artifacts

I'm wrong about Caesium - it offers a lossless option , but I didn't test it .

It didn't compress that much? How do I convert a bunch of images to lossless webp and/or avif files?
Or better yet, is there a Mac emulator for Windows so I can use ImageOptim?

[poisondeathray]

It didn't compress that much? How do I convert a bunch of images to lossless webp and/or avif files?
Or better yet, is there a Mac emulator for Windows so I can use ImageOptim?

Caesium didn't

ImageOptim didn't

Lossless webp sequence would be similar method to PNG sequence in ffmpeg using lossless=1 and output_%06d.webp . AVIF I wrote about in your other thread, currently there are issues with sequence writing

1920x1080 8bitRGB anime png (vdub2 ?)
2,211,480 bytes

Caesium PNG
2,170,199 bytes

ImageOptim PNG
2,128,576 bytes

OptiPNG
2,088,341 bytes

PNGCrush
2,088,341 bytes

PNGGauntlet
2,019,688 bytes

Acrobat PDF (jpeg2000 lossless)
1,709,341 bytes

Adobe jpeg2000 lossless
1,677,374 bytes

ImageMagick jpeg2000 lossless + IMG2PDF
1,634,776 bytes

ImageMagick jpeg2000 lossless
1,633,321 bytes

Webp lossless
1,616,644 bytes

BMF 2.01 max compression
1,542,148 bytes

JPEG-XL lossless effort7
1,325,014 bytes

JPEG-XL lossless effort9
1,292,448 bytes

EMMA 0.1.25 (Images (slow) preset)
1,059,665 bytes

All lossless image format sequences will be much larger than your original lossy YUV video that had used temporal compression.

That's the reason why your sources were so compressed in the first place - they are lossy and they used video temporal compression

Looking at different PNG compression you might shave off a few %, but it's still going to be around 35x larger than the video . It's a waste of time. PNG format and compression is old. It's like you get a new computer which is faster than your 10 year old computer. Newer algorithms like JPEG-XL will have higher compression ratios. But I would still use PNG for compatibility purposes

Any "lossless" re-encoding will decompress to uncompressed , then recompress. So filesizes for the sequence will be massive in comparison

If PNG sequence makes it 35-40x larger, JPEG-XL sequence might be 15-25x - better but still much larger than video

Nothing you do for a lossless image sequence will make it remotely close in size to the original



Lossless video will be smaller than lossless image sequences for your soruces, because it can benefit from YUV temporal compression, but still massive. Maybe 10-20x .

For ggg.mkv - 12.7MB became 447MB as an 8bit PNG sequence (which is still technically lossy, because it's integer RGB). Does it matter if you use some super slow PNG compression that might take minutes per frame to make it 440MB ? It's still a massive increase . JPEG-XL might be around 300MB (just extrapolating from the data above ) - that's still a massive increase

Lossless HEVC video re-encode became 166MB - still a massive increase, but better than any image sequence



I hope you have a good reason for re-encoding , and a better reason for using image sequences. Or start buying hard drives

[Jay123210599]

It didn't compress that much? How do I convert a bunch of images to lossless webp and/or avif files?
Or better yet, is there a Mac emulator for Windows so I can use ImageOptim?

Caesium didn't

ImageOptim didn't

Lossless webp sequence would be similar method to PNG sequence in ffmpeg using lossless=1 and output_%06d.webp . AVIF I wrote about in your other thread, currently there are issues with sequence writing

1920x1080 8bitRGB anime png (vdub2 ?)
2,211,480 bytes

Caesium PNG
2,170,199 bytes

ImageOptim PNG
2,128,576 bytes

OptiPNG
2,088,341 bytes

PNGCrush
2,088,341 bytes

PNGGauntlet
2,019,688 bytes

Acrobat PDF (jpeg2000 lossless)
1,709,341 bytes

Adobe jpeg2000 lossless
1,677,374 bytes

ImageMagick jpeg2000 lossless + IMG2PDF
1,634,776 bytes

ImageMagick jpeg2000 lossless
1,633,321 bytes

Webp lossless
1,616,644 bytes

BMF 2.01 max compression
1,542,148 bytes

JPEG-XL lossless effort7
1,325,014 bytes

JPEG-XL lossless effort9
1,292,448 bytes

EMMA 0.1.25 (Images (slow) preset)
1,059,665 bytes

All lossless image format sequences will be much larger than your original lossy YUV video that had used temporal compression.

That's the reason why your sources were so compressed in the first place - they are lossy and they used video temporal compression

Looking at different PNG compression you might shave off a few %, but it's still going to be around 35x larger than the video . It's a waste of time. PNG format and compression is old. It's like you get a new computer which is faster than your 10 year old computer. Newer algorithms like JPEG-XL will have higher compression ratios. But I would still use PNG for compatibility purposes

Any "lossless" re-encoding will decompress to uncompressed , then recompress. So filesizes for the sequence will be massive in comparison

If PNG sequence makes it 35-40x larger, JPEG-XL sequence might be 15-25x - better but still much larger than video

Nothing you do for a lossless image sequence will make it remotely close in size to the original



Lossless video will be smaller than lossless image sequences for your soruces, because it can benefit from YUV temporal compression, but still massive. Maybe 10-20x .

For ggg.mkv - 12.7MB became 447MB as an 8bit PNG sequence (which is still technically lossy, because it's integer RGB). Does it matter if you use some super slow PNG compression that might take minutes per frame to make it 440MB ? It's still a massive increase . JPEG-XL might be around 300MB (just extrapolating from the data above ) - that's still a massive increase

Lossless HEVC video re-encode became 166MB - still a massive increase, but better than any image sequence



I hope you have a good reason for re-encoding , and a better reason for using image sequences. Or start buying hard drives

All right, forget lossless formats, then. How do I make the best virtually/visually lossless image files, where the quality is so indistinguishable that even the most die-hard anime fans won't be able to tell the difference?

https://forum.videohelp.com/threads/413152-Best-Visually-Lossless-File-for-Highest-Quality

[poisondeathray]

All right, forget lossless formats, then. How do I make the best virtually/visually lossless image files, where the quality is so indistinguishable that even the most die-hard anime fans won't be able to tell the difference?

https://forum.videohelp.com/threads/413152-Best-Visually-Lossless-File-for-Highest-Quality

Nobody would download a full image sequence of anything, not even lossy.

A minimally lossy image sequence will still be massively larger than the original video

Why would someone download something 10-40x the size for similar or technically lower quality ?

A Die Hard anime fan would say you are silly for doing this .

A Die Hard anime fan would want the original video for a full sequence - much smaller filesizes, no quality loss.

Notice all the Anime fans, like the ones who made your video, use video not image sequences . None of them use image sequences - for good reasons . If you're going lossless - use video. If you're going lossy - use video. Go to some anime fan sites - can you find 1 example of an image sequence ?

But a true Die Hard anime fan would demand the true original. Not a nth generation lossy re-encode which already has artifacts


PNG is ok for a few screenshots to demonstrate something. Not good for a full sequence

[Jay123210599]

All right, forget lossless formats, then. How do I make the best virtually/visually lossless image files, where the quality is so indistinguishable that even the most die-hard anime fans won't be able to tell the difference?

https://forum.videohelp.com/threads/413152-Best-Visually-Lossless-File-for-Highest-Quality

Nobody would download a full image sequence of anything, not even lossy.

A minimally lossy image sequence will still be massively larger than the original video

Why would someone download something 10-40x the size for similar or technically lower quality ?

A Die Hard anime fan would say you are silly for doing this .

A Die Hard anime fan would want the original video for a full sequence - much smaller filesizes, no quality loss.

Notice all the Anime fans, like the ones who made your video, use video not image sequences . None of them use image sequences - for good reasons . If you're going lossless - use video. If you're going lossy - use video. Go to some anime fan sites - can you find 1 example of an image sequence ?

But a true Die Hard anime fan would demand the true original. Not a nth generation lossy re-encode which already has artifacts


PNG is ok for a few screenshots to demonstrate something. Not good for a full sequence

Well, I heard that some encodes fix problems that the original source had through filtering.

[poisondeathray]

Well, I heard that some encodes fix problems that the original source had through filtering.

Yes - But there is still no reason to use an image sequence for distribution or archive . Some workflows might use an intermediate image sequence to be later deleted - it's not left as an image sequence