Discussion Preserving quality (the scanning process and final file saving)

[Nachi]

Hello! (this is a bit lengthy post ^^;

I happen to be self though scanlaitor that 95% of the time worked as leader of the group(s) as well. So I never got the chance to learn from somebody else and only help myself with all tutorials found online. But at this point I believe this is not sufficient any more.

Since I function as leader in current (small) group I need to deal with sort of quality check and final file saving before I zip and release them.
I don't consider myself top-notch editor so quality check I do is rather mediocre.
However I would like to improve the things I can to make final product look better.

I've read many tutorials but many of them failed to explain few things for me. So I'm hoping you'll be able to do that. I know that most of it will be your personal opinion so I'd really appreciate if you'd tell what are the differences (compared to other methods) because of which you prefer to work the way you do.

1. Scanning

-single page vs. double page
I'm raw provider for my group and I do not cut books. I rarely come across a page that would really need me doing that and with a good press on the spine I can get rid of most of the gutter shadow so I think for the moment cutting books is not necessary for us.
However I've seen many times people explaining scanners should scan page by page. Why?
editor can simply cut the image in half and work on each half separately. I know that this part is not demanding but I've read that single pages make life much easier to editors. In what why?
I've also seen arguments that if you scan double page than each individual page will be to small. I've tested this only once but the size increases only if the dpi increases.

-recommended dpi
In most cases I've seen 300 and I'm using this as well (it is still faster than 400) but i also read that having a 3000+px picture is preferred by editors. However a normal manga scanned at 300 gives me a bit over 2000px high image file. Am I just confusing here recommendations for (JUMP) magazine scans vs. tankobon scans? If so what is recommended for tankobon scans (since I mostly work with those, and magazines I worked with are printed in similar quality as tankobons).

-what program to use for scanning. (with setting and saving options)
I think I'm not willing to try exporting to PS directly atm. What I have currently supports only WIA and that means I need to show the program each time how much I want scanned ie. I use to much of a time.
Since I have Cannon 4400F (will be changed in moth and half, is Canon 8800F still one of most recommended for reasonable price) I'm using the programs that came with it. Namely ArcSoft PhotoStudio and ScanGear is the interface when scanning. I don't use any filters and I presume that is still the best thing I can do, right?
But what about saving the file? Till now I've always saved in jpg at 95 compression (regardless if it's B&W or colour). I find tiff to be to big, but would using png be better? If yes, why? I know that jpg is lossy but what informations are lost that are preserved in png? What I'd like to know is how does the loss in jpg at this stage affect the appearance of final scanlated image.


2. Final file saving
I can't seem to figure out what would be the best optimum between file/size/quality.
But lets say that with nowadays connections I can afford discarding the size. what options would be the best to use for saving?

-png vs. jpg
ironically I do use png for b&w and jpg for colour pages. if using save as... I save jpg at quality 10 or 11 (depending on mood). but would something different be recommended?

-Save as vs. save for web
what's the exact difference? I'm looking into save for web options again but I don't know which one would be better. In both you can choose to save either jpg or png. I do know that for web it offers some additional options, especially for png. But those options a bit over me. If I'm using it I try to use the same settings asKalendel suggests.

-grayscale 8/16/31 channel
In all honesty this is beyond me. but I've seen somewhere someone suggesting to use 16 instead of 8 based on the raw OP presented. Would somebody mind explaining what's the ketch?

-moiré patterns (33%/66%)
I've notices that almost every time I need to deal with moiré patterns. They're not visible when picture is at full size but if it's at 33% or 66% they're visible almost all the time. Different image viewers display this differently. Some go around it, some magnify it. And I have no idea where in the process of Edited Manga can I get rid of this. I've read suggestions that you could resize first using percentage than pixels and do it step by step until moiré patterns are gone and you're at desired height/width. Since many times I'm the one doing resizing this would be handy, but it never works for me.
I know that many times filters are suggested for this but I don't want to use them. More often than not the result looks a lot uglier to me than it did before.
Any suggestions?

This last one is the one that's bugging me the most actually.
I'm presenting here and example so you can have a go as well.
clean psd (originally not scanned by me so not something I could change with different scanning method).
And png, saved for web, png-8, adaptive, diffusion, 17 colors, dither 100% (same png with dither 0%)

Any suggestions, opinions and advices would be highly appreciated

 

[Gradonil_Ral]

1. Scanning

-single page vs. double page
I'm raw provider for my group and I do not cut books. I rarely come across a page that would really need me doing that and with a good press on the spine I can get rid of most of the gutter shadow so I think for the moment cutting books is not necessary for us.

If a manga has huge enough margins on the inside of the pages, then yes, debinding is not necessary, but if they're too small or there is many double spreads, not debinding the pages will only cause more grief for the cleaner / redrawer.

Cutting a page is not necessary to debind. All you need to do is heat up the glue on the spine somehow and you can start pulling out the pages.

Spoiler: how to

There are three methods of debinding with heat (that I know of):
1. A microwave - I've never used it and I do not plan on doing so. While it might be the fastest method, it can potentially cause the most damage.
If you heat the manga for too long or too much, the pages might start to burn. Or the glue might melt entirely and spread on them.
2. A hair dryer - takes the longest and the pages can get weavy from the heated air + you can "burn" your fingers (not really, but hot air directed at your fingers can hurt a lot). On the other hand, hair dryers are easiest to come by and they won't get damaged in the process (you don't actually touch the glue with it).
3. An iron - if you don't use too much heat, it can give the best result - both time and quality-wise.
Read more here: http://hem.bredband.net/pnyxtr/scanning/iron_debind.html

I used to use the hair dryer method in the past, but I've recently switched to the iron method. TBH, I've only recently found out about it.
The thing is, you need an old iron - even if you use a sheet of paper between the manga and the iron, some glue can seep through and get onto the iron.

If you want, you can actually bind them again afterwards, you just need to get a bookbinding glue. They're not that expensive. I can't really recommend you any glue, since you live in a different country. You just need an "elastic" one - hard glues can crack with use of the pages, making them fall out.

However I've seen many times people explaining scanners should scan page by page. Why?

This is a total myth - it only comes down to the additional work you're giving the editor.

editor can simply cut the image in half and work on each half separately.

They can, but so can you. Cut the pages, that is - that's what I do. I scan doubles and then split the "false doubles" into separate pages. I leave the real doubles together - this way, an editor will not miss a double.

I know that this part is not demanding but I've read that single pages make life much easier to editors. In what why?

It might not be that demanding, but it takes time too. Doing this unnecessary manual work can totally kill editor's mood for cleaning for the day - that's what happened with me and my old raw provider. After splitting his scans into separate pages, rotating and cropping them, I didn't really feel like making the actual cleans anymore. Not for another few hours at least.

I've also seen arguments that if you scan double page than each individual page will be to small. I've tested this only once but the size increases only if the dpi increases.

Another myth. Though, perhaps it does happen with some lower quality scanners.

-recommended dpi
In most cases I've seen 300 and I'm using this as well (it is still faster than 400) but i also read that having a 3000+px picture is preferred by editors. However a normal manga scanned at 300 gives me a bit over 2000px high image file. Am I just confusing here recommendations for (JUMP) magazine scans vs. tankobon scans? If so what is recommended for tankobon scans (since I mostly work with those, and magazines I worked with are printed in similar quality as tankobons).

Nope, I like pages to be at least 3000px after cropping in whichever type of raw I'm using.
Again, this depends on a scanner - mine allows me to upscale the page without changing the dpi.
If I scan at 300dpi and 140-150% size, I'll get the 3000px high pages.

Some groups like their HQ scans to be at 600dpi. I've tried that for a volume or two. It didn't really improve anything. And since it takes a whole lot more of time to scan, I just stopped doing that.

-what program to use for scanning. (with setting and saving options)
I think I'm not willing to try exporting to PS directly atm.

I used to scan directly into Photoshop for a long time. Mainly because I wanted to split the pages before saving them. However, saving scanning, splitting and saving them one by one took too much time (the scanning software had to close and open each time), so I started scanning 20-30 pages at a time and then splitting and saving them both. Unfortunately, some times I forgot to take a break from scanning and Photoshop, having too many huge files in its memory, crashed, losing all of the recently scanned pages.
Now I just save the pages as .tiff and deal with spitting them later.

What I have currently supports only WIA and that means I need to show the program each time how much I want scanned ie. I use to much of a time.

So it doesn't remembers your settings? At all? That'd be annoying.

Since I have Cannon 4400F (will be changed in moth and half, is Canon 8800F still one of most recommended for reasonable price)

I can't say I'm familiar with any Canon devices. I had no idea that 8800F was the most recommended one.
I'm using Epson V300 Photo and I couldn't be happier. At least not about the scanning quality. The cover does not provide a good pressure to the pages, so I put a A4 book with two heavy volumes on it to make sure the pages are perfectly flat.
It's not that much of work + you do not debind, so it wouldn't affect you anyway.

I'm using the programs that came with it. Namely ArcSoft PhotoStudio and ScanGear is the interface when scanning. I don't use any filters and I presume that is still the best thing I can do, right?

I'm not familiar with these programs, but the software provided with the hardware is usually good enough.
And yes, do not add any filters.

But what about saving the file? Till now I've always saved in jpg at 95 compression (regardless if it's B&W or colour). I find tiff to be to big, but would using png be better? If yes, why? I know that jpg is lossy but what informations are lost that are preserved in png? What I'd like to know is how does the loss in jpg at this stage affect the appearance of final scanlated image.

I currently use .tiff, before that (when I imported via Photoshop) saved straight to .psd. They might take a lot of space, but how does it matter exactly? You only need to uncompress them temporarily - most of the times that can be compressed. I recommend 7-zip and .7z compression at Ultra, LZMA2 - works way better than whatever .zip or .rar can provide.
If you still prefer smaller files, then you can use .jpg. Just don't use any compression - use the maximum quality (100 / 12).
PNG is not necessary - it much bigger than .jpg and I think it was causing some troubles too. Though, perhaps it was only PNG-8 from Photoshop's "Save as web...", which my old RP thought was a good way for conserving file sizes...

Anyway, .jpg does lose some quality, but for pages of 3000px high, it doesn't really matter - it won't be noticeable in the final product.
Lossless quality does give better results when you clean, but it's really hard to notice after resizing (if at all).

2. Final file saving
I can't seem to figure out what would be the best optimum between file/size/quality.
But lets say that with nowadays connections I can afford discarding the size. what options would be the best to use for saving?

-png vs. jpg
ironically I do use png for b&w and jpg for colour pages. if using save as... I save jpg at quality 10 or 11 (depending on mood). but would something different be recommended?

PNG-8 for b&w and jpg for color, yes.

For PNG-8 I use Adaptive (no dither), 48 colors.
Why 48? So far I haven't encountered a b&w page for which that would be not enough. 6 or 32 can sometimes not be enough. 64 and more is too much.

-Save as vs. save for web
what's the exact difference? I'm looking into save for web options again but I don't know which one would be better. In both you can choose to save either jpg or png. I do know that for web it offers some additional options, especially for png. But those options a bit over me. If I'm using it I try to use the same settings asKalendel suggests.

Since "PNG-8" only exists in "Save as Web...", use it for b&w pages.
For color pages... "Save as" is enough - there's not much difference in file size between it and "Web", and "Save" distorts the images slightly less.
What quality? Depending on the raw quality and how big you want the file to get, anything between 10 and 12 is ok.

-grayscale 8/16/31 channel
In all honesty this is beyond me. but I've seen somewhere someone suggesting to use 16 instead of 8 based on the raw OP presented. Would somebody mind explaining what's the ketch?

IMO, 8-bit Grayscale is good enough.
16-bit just provides more shades of gray. Manga doesn't use that many.
You can also read this: http://askville.amazon.com/differen...ure-quality/AnswerViewer.do?requestId=1629349

-moiré patterns (33%/66%)
I've notices that almost every time I need to deal with moiré patterns. They're not visible when picture is at full size but if it's at 33% or 66% they're visible almost all the time. Different image viewers display this differently. Some go around it, some magnify it. And I have no idea where in the process of Edited Manga can I get rid of this. I've read suggestions that you could resize first using percentage than pixels and do it step by step until moiré patterns are gone and you're at desired height/width. Since many times I'm the one doing resizing this would be handy, but it never works for me.
I know that many times filters are suggested for this but I don't want to use them. More often than not the result looks a lot uglier to me than it did before.
Any suggestions?

Moi... what?
If you mean the visual effect that shows when zoom isn't at 25, 50 or 100%, then aren't they just that? A visual effect? I've never seen them at 100%, be it full size raw or the final, resized product - what do they matter?

If you mean something else, then please provide some samples. Screenshots or whatever you prefer.

This last one is the one that's bugging me the most actually.
I'm presenting here and example so you can have a go as well.
clean psd (originally not scanned by me so not something I could change with different scanning method).
And png, saved for web, png-8, adaptive, diffusion, 17 colors, dither 100% (same png with dither 0%)

I'm not exactly sure what you're asking about here.

The difference between both PNGs isn't noticeable to a naked eye. As for the difference between the PSD and the PNG, then yes, there is one.
Like I mentioned above, you 16 and 32 colors are sometimes not enough. Here, 17 colors isn't either. Try 48. It should be enough.
Also, PNG-8 darkens the pages that are in Grayscale - always change to RGB right before choosing "Save as web". The contrast won't change this way.

Oh, and... The "clean" is not 100% clean. Actually it looks a lot like a page that wasn't leveled again after denoising and saves as PNG-8 with differ, and then put into a PSD file.
If you use the white and black preview layers, you'll see the same type of noise on white and black like in the PNG files (which also aren't leveled properly), with the exception for the part where a white brush was used to remove the text.

 

[DrPepperPro]

Also, PNG-8 darkens the pages that are in Grayscale - always change to RGB right before choosing "Save as web". The contrast won't change this way.

I used to think this too and was very confused about it for a while. The actual reason is that the default color profile for grayscale PSDs in photoshop is "Dot Gain 20%", and needs to be changed to "sGray". It can be changed in edit > color settings.

 

[Gradonil_Ral]

I used to think this too and was very confused about it for a while. The actual reason is that the default color profile for grayscale PSDs in photoshop is "Dot Gain 20%", and needs to be changed to "sGray". It can be changed in edit > color settings.

This doesn't work for every shade of gray. I just tried a few saving options with both Dot Gain 20% and sGray, and, at the page I tried (light gray pattern at 1600px height + a "sample" watermark - black text at 15% opacity), the results for the patterned gray were the same as in Dot Gain 20%, but the watermark got two or three times darker.
I'd rather take my chances with RGB - there's no way of knowing which pages sGray would affect positively and which not. I have the RGB conversion in my saving action, so it doesn't take any effort from me at all.

 

[Nachi]

If a manga has huge enough margins on the inside of the pages, then yes, debinding is not necessary, but if they're too small or there is many double spreads, not debinding the pages will only cause more grief for the cleaner / redrawer.

Cutting a page is not necessary to debind. All you need to do is heat up the glue on the spine somehow and you can start pulling out the pages.

Spoiler: how to

There are three methods of debinding with heat (that I know of):
1. A microwave - I've never used it and I do not plan on doing so. While it might be the fastest method, it can potentially cause the most damage.
If you heat the manga for too long or too much, the pages might start to burn. Or the glue might melt entirely and spread on them.
2. A hair dryer - takes the longest and the pages can get weavy from the heated air + you can "burn" your fingers (not really, but hot air directed at your fingers can hurt a lot). On the other hand, hair dryers are easiest to come by and they won't get damaged in the process (you don't actually touch the glue with it).
3. An iron - if you don't use too much heat, it can give the best result - both time and quality-wise.
Read more here: http://hem.bredband.net/pnyxtr/scanning/iron_debind.html

I used to use the hair dryer method in the past, but I've recently switched to the iron method. TBH, I've only recently found out about it.
The thing is, you need an old iron - even if you use a sheet of paper between the manga and the iron, some glue can seep through and get onto the iron.

If you want, you can actually bind them again afterwards, you just need to get a bookbinding glue. They're not that expensive. I can't really recommend you any glue, since you live in a different country. You just need an "elastic" one - hard glues can crack with use of the pages, making them fall out.

I don't really have a heart to do that to books. I've debind magazine before so I know the process, but really as long as it's not necessary I don't think I'll do it ^^;;
Re-binding does worry me in case I'll ever see that necessary, but I'll be dealing with that when the time comes.

This is a total myth - it only comes down to the additional work you're giving the editor.

They can, but so can you. Cut the pages, that is - that's what I do. I scan doubles and then split the "false doubles" into separate pages. I leave the real doubles together - this way, an editor will not miss a double.

It might not be that demanding, but it takes time too. Doing this unnecessary manual work can totally kill editor's mood for cleaning for the day - that's what happened with me and my old raw provider. After splitting his scans into separate pages, rotating and cropping them, I didn't really feel like making the actual cleans anymore. Not for another few hours at least.

Another myth. Though, perhaps it does happen with some lower quality scanners.

Good to know most of it is a myth ^^ But definitely point take that I could split up the pages.

Nope, I like pages to be at least 3000px after cropping in whichever type of raw I'm using.
Again, this depends on a scanner - mine allows me to upscale the page without changing the dpi.
If I scan at 300dpi and 140-150% size, I'll get the 3000px high pages.

Some groups like their HQ scans to be at 600dpi. I've tried that for a volume or two. It didn't really improve anything. And since it takes a whole lot more of time to scan, I just stopped doing that.

Unfortunately what I'm currently using doesn't allow me upscaling :/ or at least I haven found any function that would enable this.
And 600dpi for anything else than colour probably is just a waste of time ^^

I used to scan directly into Photoshop for a long time. Mainly because I wanted to split the pages before saving them. However, saving scanning, splitting and saving them one by one took too much time (the scanning software had to close and open each time), so I started scanning 20-30 pages at a time and then splitting and saving them both. Unfortunately, some times I forgot to take a break from scanning and Photoshop, having too many huge files in its memory, crashed, losing all of the recently scanned pages.
Now I just save the pages as .tiff and deal with spitting them later.


So it doesn't remembers your settings? At all? That'd be annoying.

No it doesn't save the settings, that's why I'm working with the programs I got with the scanner.

I can't say I'm familiar with any Canon devices. I had no idea that 8800F was the most recommended one.
I'm using Epson V300 Photo and I couldn't be happier. At least not about the scanning quality. The cover does not provide a good pressure to the pages, so I put a A4 book with two heavy volumes on it to make sure the pages are perfectly flat.
It's not that much of work + you do not debind, so it wouldn't affect you anyway.

Actually I bought my scanner about 7 years ago so it really is old and It was more of a coincidence I choose the right one (at the time 4400F was most often recommended) as for 8800F I'm just asking based on topics in this section that are dealing with scanning, but I think the newest is from 2010 or something winch is pretty old in itself as well. I'll take a look at Epson too, I like the idea of being able to upscal the image.

I currently use .tiff, before that (when I imported via Photoshop) saved straight to .psd. They might take a lot of space, but how does it matter exactly? You only need to uncompress them temporarily - most of the times that can be compressed. I recommend 7-zip and .7z compression at Ultra, LZMA2 - works way better than whatever .zip or .rar can provide.
If you still prefer smaller files, then you can use .jpg. Just don't use any compression - use the maximum quality (100 / 12).
PNG is not necessary - it much bigger than .jpg and I think it was causing some troubles too. Though, perhaps it was only PNG-8 from Photoshop's "Save as web...", which my old RP thought was a good way for conserving file sizes...

Anyway, .jpg does lose some quality, but for pages of 3000px high, it doesn't really matter - it won't be noticeable in the final product.
Lossless quality does give better results when you clean, but it's really hard to notice after resizing (if at all).

I'm not that familiar with tiff so I'm hesitant to use it. Huge size doesn't help either, thanks to that I don't know how compressible the file is, and the final size (for raw) of zip/rar/7z is important to since my connection is awfully slow. But I guess it wouldn't hurt experimenting a bit

For PNG-8 I use Adaptive (no dither), 48 colors.
Why 48? So far I haven't encountered a b&w page for which that would be not enough. 6 or 32 can sometimes not be enough. 64 and more is too much.

Thank you! I always though something wrong with the final picture and I tried with 32 colours but something was still off, I just never knew what. Hopefully this will solve my problem

Moi... what?
If you mean the visual effect that shows when zoom isn't at 25, 50 or 100%, then aren't they just that? A visual effect? I've never seen them at 100%, be it full size raw or the final, resized product - what do they matter?

If you mean something else, then please provide some samples. Screenshots or whatever you prefer.

It's when patterns start to overlap and create new ones even though they're not really there. Google gives you many examples but in manga its a bit different. Essentially it's what you said the effect when the zoom isn't at certain percentage. For a while now I'm suspecting this depends on the size of the image, but than again I've seen releases where these patterns are not present no matter at what percentage the zoom is while others have it. I know that different image viewers present this differently but using default Windows viewer (which I presume most readers use) it looks like this:

Spoiler

It's most blatantly obvious on second panel. Of course when put into slide show its different, but on second panel is still easily noticeable:

Spoiler

I just wondered if its possible to get rid of this and not ruin the actual patterns?

Oh, and... The "clean" is not 100% clean. Actually it looks a lot like a page that wasn't leveled again after denoising and saves as PNG-8 with differ, and then put into a PSD file.
If you use the white and black preview layers, you'll see the same type of noise on white and black like in the PNG files (which also aren't levelled properly), with the exception for the part where a white brush was used to remove the text.

With "clean" i meant no text ^^;; but thank you for reminding me of contrast... I usually do this with levels only and sometimes over level by point or two :/
The particular picture actually went through many levelling already since it was not scanned by me. The scanner in this case scanned for public and done some very basic crop/straighten/level job often making an editors job a nightmare. (cause the page is not straight as well but rotating any more would do more damage than good in this case.)



And the sGray appears to be hit or miss. Good to know.

 

[Gradonil_Ral]

It's when patterns start to overlap and create new ones even though they're not really there. Google gives you many examples but in manga its a bit different. Essentially it's what you said the effect when the zoom isn't at certain percentage. For a while now I'm suspecting this depends on the size of the image, but than again I've seen releases where these patterns are not present no matter at what percentage the zoom is while others have it. I know that different image viewers present this differently but using default Windows viewer (which I presume most readers use) it looks like this:

Spoiler

Oh, that's what you meant. I really didn't see how they would appear in the final product. Now I know.
Well, Windows Photo Viewer was meant as a quick way for normal people to view their photos. It was certainly not meant for viewing manga xD Its resizing script is poorly made.
You should just stop using it. Really. There's no point.
I recommend "CDisplay" (not the EX version|). It's the easiest/smallest manga reading program that I know of. It can read the images that you load into it or the images that are in the zip/rar archives you choose. You can resize the pages in it and no moire patterns will appear. You can read through the entire chapter just bu using scroll in your mouse (or you can change the mouse settings to act the way you want). You can even change color balance, etc. - I use it for older releases where pages haven't been leveled properly.
And what I like the most, is that it opens on full screen with no menus whatsoever - nothing to distract you from reading (if you prefer a normal window, you can change to that too).
In any case, it has all the necessary options.

And no, Windows Viewer is most the most used option for reading manga - online reader are. Most of people that aren't too lazy to download manga are also not too lazy to find an offline manga reader that work for their needs.

I just wondered if its possible to get rid of this and not ruin the actual patterns?

I think that the moire patterns might not appear depending on the height of the page, its sharpness and contrast (how much you leveled). Also I've tested a few types of releases just now, even my own cleans, and it seems that they appear most often on denoised pages. Topaz does clean up the real patterns nicely, but it also slightly reshapes them - if you resized a raw to the same size as a cleaned page and compared them both, you wouldn't probably see any moire patters on the raw. Or, if you did, they would be far less noticeable than on the clean.

The particular picture actually went through many levelling already since it was not scanned by me. The scanner in this case scanned for public and done some very basic crop/straighten/level job often making an editors job a nightmare. (cause the page is not straight as well but rotating any more would do more damage than good in this case.)

Even if the raws came pre-leveled, if you use any filters - be it topaz filters, surface blur, or another way of denoising - you need to always slightly level afterwards. The adjustment layers from the video I linked in my previous post show all leftover dirt/noise - they can be useful for choosing the settings for the first leveling, or the last one. I use them each time I level.
I also use them when I burn/dodge the leftover noise/dirt out.

 

[DrPepperPro]

This doesn't work for every shade of gray. I just tried a few saving options with both Dot Gain 20% and sGray, and, at the page I tried (light gray pattern at 1600px height + a "sample" watermark - black text at 15% opacity), the results for the patterned gray were the same as in Dot Gain 20%, but the watermark got two or three times darker.
I'd rather take my chances with RGB - there's no way of knowing which pages sGray would affect positively and which not. I have the RGB conversion in my saving action, so it doesn't take any effort from me at all.

Well converting a scan (which most likely isn't dotgrain 20% in the first place) to dot grain 20% is what darkens the grays, so some colors will be lost. It doesn't show up in photoshop unless you use proof colors (ctrl + y). Using save for web doesn't preserve color profiles so an image with the standard 8bit 256 grays will look the same whether saved from 8bit sRGB or sGray.

I also used to have a merge all and convert to sRGB in my saving action too, but it was annoying when I sometimes needed to save manually.

 

[Gradonil_Ral]

Merge all? Why would you merge all?

My action (for saving gray pages) goes like this:
1. Convert to RGB (without flatten).
2. Export (Save as Web: PNG-8 Adaptive 48 colors)
3. Convert to Grayscale (without flatten).
4. Save (the psd).
5. Next document.

And my batch action just plays it 19 times (the number of pages most chapters have -> 20 with one double = 19).

 

[DrPepperPro]

Oh, merge all just to make sure levels layers don't change with the profile change.

 

[Gradonil_Ral]

Oh, I merge down the levels layers as I go. There's no need to merge the rest.

 

[Nachi]

Oh, that's what you meant. I really didn't see how they would appear in the final product. Now I know.
Well, Windows Photo Viewer was meant as a quick way for normal people to view their photos. It was certainly not meant for viewing manga xD Its resizing script is poorly made.
You should just stop using it. Really. There's no point.
I recommend "CDisplay" (not the EX version|). It's the easiest/smallest manga reading program that I know of. It can read the images that you load into it or the images that are in the zip/rar archives you choose. You can resize the pages in it and no moire patterns will appear. You can read through the entire chapter just bu using scroll in your mouse (or you can change the mouse settings to act the way you want). You can even change color balance, etc. - I use it for older releases where pages haven't been leveled properly.
And what I like the most, is that it opens on full screen with no menus whatsoever - nothing to distract you from reading (if you prefer a normal window, you can change to that too).
In any case, it has all the necessary options.


And no, Windows Viewer is most the most used option for reading manga - online reader are. Most of people that aren't too lazy to download manga are also not too lazy to find an offline manga reader that work for their needs.

Well even if we discard the online readers ^^ I guess I don't fit to the normal since I've been using the default that system has been offering to me all these years. I don't remember having any major problems with Ubuntu, but Windows hasn't changed for last 10 years with this one :/
I tried CDisplay though and while it reduces moire greatly it's still there... just a bit though ^^

I think that the moire patterns might not appear depending on the height of the page, its sharpness and contrast (how much you leveled). Also I've tested a few types of releases just now, even my own cleans, and it seems that they appear most often on denoised pages. Topaz does clean up the real patterns nicely, but it also slightly reshapes them - if you resized a raw to the same size as a cleaned page and compared them both, you wouldn't probably see any moire patters on the raw. Or, if you did, they would be far less noticeable than on the clean.

Yeah, I think it's on me now to play with size and contrast now. The ketch is that I don't use any denoise filters at all. 99% of the time I work with tankobon scans anyway and using filters on them just makes things worse. That's why I always found it strange.
As you said raw at same size as clean page doesn't show them. Lol.. actually now that I think about it the answer was always right in front of me xD



I remembered one more thing... what's better and why. Making final resize before typesetting or after? Editors send me both depending on how they were taught in the past...

 

[Brandan White]

I face problem with File Saving Process because tutorial for File Saving give me bad results loss screen tone effects i.e pattern got something and pattern loss its details. I can't put it in word because i don't know what should i call it so i use 256 colour which increase file size but give good data.

I worked on Digital Copy of Volumes and Digital Version of Magazine for Chapter.

Note:This Thread was difficult to find in search

 

[Mana Light]

Just release it at png - 8, 256 color. (png 24 or jpg for color page)
For cleans from mag scans, I see 16 color or 32 color is enough. Normally other groups just release it at that color. However my cleans from mag raws have extremely many details so it needs 256 color to look good. I release at 256 color as well.
And I see the different between 256 color with lower color can be seen easily with size >= 1600 px high. Digital raws like Kodansha (1600 px) needs to released at png-8, 256 color too.
PNG-24 maybe is better but file size is much bigger so don't. I use PNG-24 for color page instead of jpg.
Who cares about other people's internet speed. Nowadays, internet speed is fast too so although file size is bigger compare to 16 or 32 color, imo 256 color is acceptable.

 

[mantou]

I know there's a similar thread but I'm still confused. What is the best recommended way to save the files as png after qc-ing? I'm really puzzled about this as it seems that everyone has their own way of saving. I'm currently following this (x). How do you save the files such that quality is not compromised and the scans are not altered in terms of colour too?
Also hoping to know for saving in jpg mode too.

I have never tried to save files in tiff only psd but what's the setting if you save the files in tiff mode?

(BTW, this is a dumb question but the file size of scans determines their quality, right? Are HQ scans usually more than 1 MB big?)

 

[Brandan White]

For Colour pages we save them in jpg for black and white we use png

File Size depend on Image Quality like some pages need 8 bit colour to 256 depend on details sometime details were lost like pattern lost details etc

 

[mantou]

For Colour pages we save them in jpg for black and white we use png

File Size depend on Image Quality like some pages need 8 bit colour to 256 depend on details sometime details were lost like pattern lost details etc

ty! do you use save as or save for web? if it's save for web, what's the recommended setting?

 

[Gradonil_Ral]

PNG-8 Is only avaliable in Save as web, so yes.

The settings depend on whether you wanna save up space or not. Although most pages won't lose quality with PNG-8 Adaptive No-Differ 32 Colors, there might be some that will. Usually 64 colors should handle all. But if you wanna save up on space I recommend 48 colors. And if you really don't care about saving space, just go with 256 colors, like @Brandan White suggested.
Just remember to either 1 - always convert to RGB before saving B&W pages with save as web (otherwise the patterns will darken) or 2 - read what @DrPepperPro wrote in post #3 in here (but like I said in post #4, it doesn't always work, IMO).

For color pages save in jpg, either as web or normally. 80-100% (10-12) quality. I usually use max.

 

[mantou]

PNG-8 Is only avaliable in Save as web, so yes.

The settings depend on whether you wanna save up space or not. Although most pages won't lose quality with PNG-8 Adaptive No-Differ 32 Colors, there might be some that will. Usually 64 colors should handle all. But if you wanna save up on space I recommend 48 colors. And if you really don't care about saving space, just go with 256 colors, like @Brandan White suggested.
Just remember to either 1 - always convert to RGB before saving B&W pages with save as web (otherwise the patterns will darken) or 2 - read what @DrPepperPro wrote in post #3 in here (but like I said in post #4, it doesn't always work, IMO).

For color pages save in jpg, either as web or normally. 80-100% (10-12) quality. I usually use max.

thanks for the clarification! i tried converting to rgb before saving but the leveling gets affected instead. (for example, i levelled the blacks black but after converting, some of the lighter spots come back again and i have to level it again. how do you convert it so that everything remains the same except the lightening of patterns? or do you do the levelling in rgb mode? or flatten everything before converting?

 

[Gradonil_Ral]

The only way leveling could be affected due to color mode change is if you use adjustment layers to level.
It is alright to use adjustment layers, but you need to merge them down afterwards. Otherwise the image is leveled "virtually". Meaning, YOU see the layer as leveled, but the program doesn't. Using denoise or other filters, or even burn/dodge won't give you the effects you want.
If you kept levels as adjustment layers because you were worried you might've chosen incorrect settings and thought it was easier to adjust them this way, then I suggest that you should instead duplicate the layer you edit to make it easier to "go back" to the previous state. It is more effective then going back in history states, which, btw, disappear too quickly if you're doing any manual work (burn/dodge or brush).

TL;DR:
Merge down adjustment (levels) layers before converting modes.