Ren'Py What's considered the standard file size for images?

[Deleted member 5906128]

Currently averaging around 3 MB per image, which adds up REAL fast.



I assume it's standard practice to compress images first when looking to lower overall file size, but by how much?

 

[BzPz]

Use

You must be registered to see the links

and play around with the settings to see what results you get, too much reduction and you'll lose image quality.
Just converting to JPG at around 95% quality nets you a very good reduction.

Find what is acceptable for you, everyone has different standards.

 

[MissFortune]

Nearly anything but PNG, honestly.

Lossless WebP at 1080p is basically around 1MB or so and tends to retain nearly all of the native quality.

 

[GNVE]

I get great results with

You must be registered to see the links

. It costs a little money ($25 a year I believe?) if you want to do many images at once or you need a larger file size. But from the hundreds if not thousands of images I processed only 1 fucked up. I have in the past compared the original to the compressed image zoomed in at 200% or so and could not find any differences in quality.
It shrinks file sizes by 65% - 75% for PNG images. It does JPEG and WebP as well and seems to still save 10% - 30% based on one quick test. So that would imply that PNG is about 2 - 2,5x larger than WebP with this tool at around 1MB.

 

[79flavors]

While I don't develop, I'd render the biggest .PNG files I had the patience to wait for the render time, then keep them in one folder.
Then I'd mass convert anything in that folder to another folder, then use the smaller images for RenPy.
It's much easier to convert images to be smaller, using larger images - both in terms of resolution and image quality.

For anything that requires transparency (sprites, buttons, imagemaps, etc)... you'll need to use either .png or .webp.
For everything else (backgrounds, etc)... you can use .jpg or .webp.

.webp images are generally smaller and are the newer technology, but being the ludite I am - I'd probably use .jpg and .png.

Don't get misled into thinking that one format is automatically smaller file sizes than another. A lot of it is down to the compression and image quality settings. It's easily possible to have a high quality .jpg image be larger than the same low quality .PNG image. A lot of the "mass convert" tools default to higher compression/lower image quality - which can lead to people thinking the other file formats are significantly better. Don't get me wrong, some file formats ARE inherently smaller - I'm just saying the file format is not the ONLY factor (or in some cases, even the most impactful).

My approach would be to convert to the lowest quality .jpg files I think I can cope with (where 90%+ of the converted images look okay to my eye). Then for the remaining 10%, manually re-convert them on incrementally higher settings until I'm happy to use that image. For the images that need transparency, I'd do them separately to .png, adopting the same "go small, but increase quality as needed".

The thing you need to keep an eye on is renders with large areas of uniform color. Light blue walls/sky jump to mind, but same for black trousers/skirts or blue jeans. On low quality/high compression - these will lead to areas of the picture that should be one solid color - but looks like circles within circles of gradually changing colors. If you see that, just up the image quality a bit and try again.

 

[Deleted member 5906128]

Lots of great answers here, thanks guys!

 

[anne O'nymous]

Don't get misled into thinking that one format is automatically smaller file sizes than another.

And don't get stuck with one and only one compression factor because:

The thing you need to keep an eye on is renders with large areas of uniform color. Light blue walls/sky jump to mind, but same for black trousers/skirts or blue jeans. On low quality/high compression - these will lead to areas of the picture that should be one solid color - but looks like circles within circles of gradually changing colors. If you see that, just up the image quality a bit and try again.

Some renders will tolerate heavy compression, others would already be limit with a light one.

It imply more works, but starting with an average compression factor, then increasing/decreasing render by render is the way to reach a satisfying compromise between size and quality.

 

[immortalkid69]

While I don't develop, I'd render the biggest .PNG files I had the patience to wait for the render time, then keep them in one folder.
Then I'd mass convert anything in that folder to another folder, then use the smaller images for RenPy.
It's much easier to convert images to be smaller, using larger images - both in terms of resolution and image quality.

For anything that requires transparency (sprites, buttons, imagemaps, etc)... you'll need to use either .png or .webp.
For everything else (backgrounds, etc)... you can use .jpg or .webp.

.webp images are generally smaller and are the newer technology, but being the ludite I am - I'd probably use .jpg and .png.

Don't get misled into thinking that one format is automatically smaller file sizes than another. A lot of it is down to the compression and image quality settings. It's easily possible to have a high quality .jpg image be larger than the same low quality .PNG image. A lot of the "mass convert" tools default to higher compression/lower image quality - which can lead to people thinking the other file formats are significantly better. Don't get me wrong, some file formats ARE inherently smaller - I'm just saying the file format is not the ONLY factor (or in some cases, even the most impactful).

My approach would be to convert to the lowest quality .jpg files I think I can cope with (where 90%+ of the converted images look okay to my eye). Then for the remaining 10%, manually re-convert them on incrementally higher settings until I'm happy to use that image. For the images that need transparency, I'd do them separately to .png, adopting the same "go small, but increase quality as needed".

The thing you need to keep an eye on is renders with large areas of uniform color. Light blue walls/sky jump to mind, but same for black trousers/skirts or blue jeans. On low quality/high compression - these will lead to areas of the picture that should be one solid color - but looks like circles within circles of gradually changing colors. If you see that, just up the image quality a bit and try again.

what software would you use to mass convert daz png render files?

 

[MissFortune]

what software would you use to mass convert daz png render files?

XnConvert is free, and basically can do anything you need (conversion-wise) en masse. Not too sure what the paid options out there are or if they're any better than XnConvert, though.

 

[anne O'nymous]

XnConvert is free, and basically can do anything you need (conversion-wise) en masse. Not too sure what the paid options out there are or if they're any better than XnConvert, though.

A bit more complex to use, but more powerful, there's

You must be registered to see the links

.

It's a command-line tool, hence the "more complex", but fully dedicated to image manipulations. It need a bit of learning (or to ask google), but once you wrote few batch files, a big part of the post processing can be automatically done. It will not always give the expected result, but the few time it failed can still be manually processed by GIMP/Photoshop/whatever, while the times it worked were time free for you.

By example, with this:

@ECHO OFF
DIR /b *.png > magick.lst
FOR /F "tokens=1,2 delims=." %%f IN ( magick.lst ) DO (
    FOR /L %%c IN (94,-1,75) DO magick "%%f.%%g" -quality %%c%% "%%f-%%c.%%g"
)
DEL magick.lst

For each PNG files in a directory, you get 20 JPEG files with a compression factor going from 94 to 75. A bit useless, but it's the only example I have under the hand.
The same king of batch file can be used to generate WEBP files with 3-5 different compression factors, letting you keep the one that looks the best. And in the same time it can also take care of the resizing if you render at higher resolution than what you'll use in your game.

Composing an image based on different renders (when there's a crowd by example) is also something relatively easy to automatize, and so on. It will not replace too complex filters, but most of the basic post processing task (gamma correction, contrast adjustment, etc) can be done with it.
The only problem is what I said, it need a learning time, for both imageMagick and DOS batch files, before you can effectively use it. But google can help for this, and once you've the batch files it can proceed a day of renders in less time that what you would need to manually proceed one with GIMP/Photoshop/whatever.

But, and I insist on this point, it need a learning time. What mean that there's a benefit only if you learn before you start your game. Else the time you'll pass learning risk to be higher than the time you'll gain on the post processing. Not because it's difficult to use it by itself, but for complex manipulations, like contrast adjustment by example, it will need a lot of trial and error before you find the value combination that you need to use. And you'll also need to have more than one of them because not all renders will need the same level of correction.

 

[immortalkid69]

A bit more complex to use, but more powerful, there's

You must be registered to see the links

.

It's a command-line tool, hence the "more complex", but fully dedicated to image manipulations. It need a bit of learning (or to ask google), but once you wrote few batch files, a big part of the post processing can be automatically done. It will not always give the expected result, but the few time it failed can still be manually processed by GIMP/Photoshop/whatever, while the times it worked were time free for you.

By example, with this:

@ECHO OFF
DIR /b *.png > magick.lst
FOR /F "tokens=1,2 delims=." %%f IN ( magick.lst ) DO (
    FOR /L %%c IN (94,-1,75) DO magick "%%f.%%g" -quality %%c%% "%%f-%%c.%%g"
)
DEL magick.lst

For each PNG files in a directory, you get 20 JPEG files with a compression factor going from 94 to 75. A bit useless, but it's the only example I have under the hand.
The same king of batch file can be used to generate WEBP files with 3-5 different compression factors, letting you keep the one that looks the best. And in the same time it can also take care of the resizing if you render at higher resolution than what you'll use in your game.

Composing an image based on different renders (when there's a crowd by example) is also something relatively easy to automatize, and so on. It will not replace too complex filters, but most of the basic post processing task (gamma correction, contrast adjustment, etc) can be done with it.
The only problem is what I said, it need a learning time, for both imageMagick and DOS batch files, before you can effectively use it. But google can help for this, and once you've the batch files it can proceed a day of renders in less time that what you would need to manually proceed one with GIMP/Photoshop/whatever.

But, and I insist on this point, it need a learning time. What mean that there's a benefit only if you learn before you start your game. Else the time you'll pass learning risk to be higher than the time you'll gain on the post processing. Not because it's difficult to use it by itself, but for complex manipulations, like contrast adjustment by example, it will need a lot of trial and error before you find the value combination that you need to use. And you'll also need to have more than one of them because not all renders will need the same level of correction.

yeah seems like this complex part could be a turn off for a lot of guys who are just looking for a quick way to convert their images. But I guess it is good to know about this.