Raw: /g/3132456/e9e245e35a/

Ah, otaku love.........good for them :D

A great one, thanks for all the effort and sharing to all involved, nice work :)

Base +57, SkullDraecos +6, BonoBono08 +6

Great translation + typesetting.
PSA: You can convert B&W pages to 8 bit greyscale PNG for no loss in quality, and usually a much lower size than JPG if the source is clean. This gallery went down by 28 MB when doing so, but could've been more like 42 MB if not for the compression artifacts. A tool like OptiPNG, or ECT can be used afterwards to optimize it more.
Also, lorem ipsum on pg. 21 top right.

Base +9, Azriel_Satan +6, West Bird +6

Girls like her are called landmine girls for a reason.

Base +16, BonoBono08 +6

Vanilla Impregnation Actually Exists! Praise The Lord!

Base +6, lolsaucerszd +8, Piter993 +9, BonoBono08 +6, ugge123 +11, EisakuOtani +10

hope someone will decensor these small but annoying bars
@rfbna, i wouldn't suggest to compress or optimize files in any way, since if not done correctly it could lead to banding and other nasty things, every time you decrease size and delete bits there is a consequence, sometime the files you work on might have already lost many bits compared to the real source. So its better to post the files the more untouched as possible, then people that care and know about it do it for themselves. But i can see this site having an alternative button download for people that want to save space and don't know the process, where someone went for each one and found the best compression for png or even newer formats that while lossy still save much more space and are visually identical. I do it for myself since by this rate in some years, keeping original quality will have a big impact.

Base +6

Always put a baby in crazy.

Base +6

@dafdagh That's really only a concern with lossy formats like JPG which PNG is not, I've never seen any degradation when converting an image to PNG except for some specific lossy webp images which thankfully none of these are. There is the chance for user error though if a seemingly B&W page actually does have some color in it, in that case you could just batch export to PNG without using any modifiers like greyscale, and then let ECT handle it using the --strict parameter to ensure absolutely nothing can go wrong. If you do have any examples of an image that can't convert to PNG without issues I'd appreciate it if you could share so I could try replicating it.

Last edited on 24 December 2024, 05:24.

Base +9

@rfbna when something is as clean as this you wont have any problem, but when there any errors and artifacts trying to save space without first fixing them will only worsen the result. For example banding can increase when changing color mode and even saving to png/lossless doesn't guarantee that no problem will occur. So yea my point is that you can save space when you are sure everything is perfect, otherwise don't change color mode or format or file size. That's why i don't agree to recommend the same method for every case

p.s if you have a good gui to compress tell me as im trying new formats and changing programs often to find the most user friendly while powerful one

Last edited on 24 December 2024, 09:16.

Base +6

@dafdagh I see, I didn't mean to imply that they should be modified further if they aren't already clean in the first place, I'll amend that going forward. I'm still standing by the last method I mentioned though, even as an all case scenario.

If you're trying new formats for personal use, JXL is the best for lossless compression outside of edge cases like webp on pixel art. I personally use command line so I'm not familiar with gui solutions, but I've heard this is good https://github.com/JacobDev1/xl-converter . When compressing JPG there's essentially 3 modes you have to know about, 1. VarDCT which is only used on lossy operations 2. Modular which is used to losslessly compress other formats besides JPG using raw pixels, but can be used to compress JPG as well if you specify it. This on average compresses down by 35% on PNG. 3. lossless transcoding which is the special mode exclusive to JXL, and is on by default. It essentially reuses the information that JPG used to compress in the first place. This mode can be safely used on everything since it also lets you get back the original JPG with the exact same file hash, it on average compresses by 20%. Only issue with this is that it's too faithful to the 20% mark, and even an obviously bloated JPG won't compress down any further.
Using this gallery as an example, compress the color JPGs using the special mode for 20% off with an effort of 7-9, and then the B&W ones using the modular mode. At an effort value of 7-9, the B&W pages go down to 49.6-51.9 MB. There are instances you can use modular on color JPGs, but only if they were saved at like 100q, and the image is sharp + flat color wise. End result is 100.5 > 52.0 at e9, there's an e10 which is 10% smaller than e9, but takes ages to encode on modular, and isn't worth using on special.
If you're fine with lossy though, and q90/d1 doesn't compress well enough for you on JXL, just use AVIF by that point unless you value sharper details at the cost of compression artifact noise. Do note though that some things surprisingly compress better lossless like clean B&W halftone pages.

Base +9

Indeed i started using that program when i first tried jxl. For example using q90 on page 12 went down to 1.3mb, so more than half. And checking modular mode down to 1mb without losing much, just a little smoother. But the program lack many things including color modes and tones... anyway im not sure jxl can bring this gallery down to 30 mb, more than 3 times, without changing tones or going lossy.
As for the halftones i like to remove them with some spices, as they are annoying to watch, and given they contain much less they only inflate resolution/space.

Base +6

@dafdagh If you're noticing ANY loss at all, then that means it's not lossless. I'm guessing the gui program doesn't allow you to use lossless modular directly on JPG since page 12 only went down to 1.4 MB for me even at effort 10 while it was 1.6 MB at effort 9. You can work around this by converting them to PNG first, and making sure the lossless mode is ticked so it uses lossless modular instead. This gallery technically can go down to 33.2 MB while being lossless, but only on the raw since this gallery is JPGified a bit compared to it. If a gallery is exported as JPG while being smaller, or even the same size as the original raw, you can bet there's going to be extra artifacting introduced that ensures it won't compress as well as the original. This is half of the reason why I want to see PNG being used at all times for B&W unless the source being worked on is already so compressed that using PNG would blow up the size by 10x, or something.
Regarding lossy, there's no benefit to re encoding the same image multiple times since you're just adding extra generational loss on top that will usually make it even harder to compress. AVIF is definitely the way to go at this point if you're aiming to go that low in file size for this specific gallery, here's an example of AVIF at exactly 30.0 MB https://pixeldrain.com/u/yhyAptfm pixeldrain reports it being a bit larger than it actually is.
When it comes to actual editing of images for the purpose of compression, there's no gui program that I know of besides just straight up using an actual image editor like photoshop unfortunately.
Hope this clears everything up, and Merry Christmas.

Base +9

I usually use q90 for jxl as its considered the visually lossless setting, and indeed you need big zooms to notice a difference. But really i dont use mega zooms and when looking from a tablet even q90 is very conservative. But yes i agree that when working and fixing stuff from the pc its simply reasonable to do so in a lossless chamber, compression is just the final step. I tried Avif a little bit, but as it wasnt built for image fidelity i ditched it for jxl, which i think treat noise and banding greatly when going lossy, infact its the only one to do so every time especially with presence of large tonal ranges which automatically make any other go wonky blocky on lossy and even lossless ( webp )
From what i understand you always use lossless but do you actually fix the errors at any point? my idea is that its perfectly fine to go lossy once i clean all, as long as i dont introduce or notice problems (banding, halos, aliasing and so on). Its basically the same logic of encoding animated series.
I guess you like to compress while staying lossless while i compress to clean and save space.
Anyway happy holidays :)

Base +6