Please add dither for reading JPEG files

[Xn User]

Hi )

I for a long time not known how to say it) , and because english not my native languarge , I can not expressing my idea ) I create topic viewtopic.php?f=79&t=48281 and viewtopic.php?p=202804 where I saying about bad quality in JPEG files if use :
formula subsampling 2x2,1x1,1x1
or
subsampling 2x1,1x1,1x1
or
subsampling 1x2,1x1,1x1
or any which difer from formula subsampling 1x1,1x1,1x1
in this mode image low quality and rude colors , please add dither for JPEG images as this make ACDSee Pro with IDE_ACDSTD.APL version 5.0 and newer )
https://en.wikipedia.org/wiki/Dither
not need read JPEG In the frontal without dithering ) because will be rude gradients of colors

[user0]

antialiasing filter at 100% zoom?

[Xn User]

Yes ) I say about with dither JPEG reading always ) you can compare color counts from here https://app.mediafire.com/xs60z866xbauo and here https://www.mediafire.com/folder/6b9mczp3re4ea/Sample and you can see what ACDSee Pro with IDE_ACDSTD.apl 5.0 always in decode process dither and color count more more more higher when XnView MP , this do will more smoothing images JPEG if use sub sampling different from 1x1,1x1,1x1 ) this is still necessary for transcoding images with high quality

[Xn User]

By the way dither can add as optional as option in reading JPEG format , and default can set as disable , and if user want , so can enable dither and set value level from 0 to 100 )

[Xn User]

Please add to setting as read this:
https://www.mediafire.com/view/68935toe ... 0.png/file
https://www.mediafire.com/view/yioofdmt ... 9.png/file
only TrueColor images in JPEG tab, I think, what Floyd-Steinberg will be sufficient for improve quality reading JPEG files with 2x1 1x1 1x1 or 1x2 1x1 1x1 or 2x2 1x1 1x1.

[user0]

You are confused.

ACDSee and XnViewMP decodes JPEG in exactly the same way, results match 100% when you compare:
- JPEG resaved in lossless format
- JPEG screengrabbed at 100%

There will be difference only when comparing screenshots of zoomed-in/out image, as image scaling algorithms (aka interpolation, resample methods, resize filters, antialiasing[edge-smoothing during interpolation]) + enhancements are applied.

Check apps default settings:

• XnViewMP (Settings > View)
  • interpolation
    Zoom-in [none]
    Zoom-out [Bilinear]
  • enhancements
    ☐ Use sharpen

• ACDSee (Tools > Options > View mode)
  • interpolation
    ☐ Show pixels when zoomed past 100% - unknown algorithm is applied for zoom-in
    Zoom-out - no info which algorithm is used
  • enhancements
    ☑ Sharpen subsampled images - unknown sharpen % is applied

[Xn User]

I'm sorry , but I think what you all not understanding and not want understand what I write (
You can understanding differ between zooming and dithering ?
As I understand , you still think what I sayding about zoom process , this but your error , I'm sayding about quality encoding , and because JPEG encoding image do rude colors for smaller file size , that's why need do dithering colors in process decoding for improve quality JPEG file
I for that and give files with original size for you can compare result ACDSee and XnView MP
https://app.mediafire.com/6b9mczp3re4ea
this is files was recived on methods , open original sample.jpg in ACDSee Pro 5 with ide_acdstd.apl version 5.0 and save as png , next open original sample.jpg in XnView MP and save as png , and see result

I'm will try explain again...
We have image :

We compress image to JPEG and JPEG do this :

Next we open result image , and XnView read as is , I think what tihis rude , very rude colors , that's why I offer apply dither in decoding process to get result :

this possible do ?

[user0]

There is NO ANY DIFFERENCE in your originals in ACDSee 2025 and XnViewMP at 100% (with default settings).
Use tools that can show diff, eg Beyond Compare, black color means 100% match (in screenshot or resaved).

Your problem is ACDSee 5 from Feb 14, 2008.
You need to complain at ACDSee forum that devs bRoKe JPEG decoder, not here

[Xn User]

Your problem is ACDSee 5 from Feb 14, 2008.
You need to complain at ACDSee forum that devs bRoKe JPEG decoder, not here

Why? You realy think what if images in ACDSee Pro 5.0 with ide_acdstd.apl version 5.0 look better , so me need to complain on this ? You sure?

[Xn User]

I'm understand what you do... you open original file and file recived from XnView MP , if yes , so clear why you so sayd
But this not means what so better

[user0]

I'm understand what you do... you open original file and file recived from XnView MP , if yes , so clear why you so sayd
But this not means what so better

as I explained in previous post, I used your originals (original.jpg and original sample.jpg) in both apps,
then resaved as png for compare1 and made screenshots for compare2 (on pics). Results are the same - 100% match.

images in ACDSee Pro 5.0 with ide_acdstd.apl version 5.0 look better

sure, but you should not keep this secret here, you need to tell ACDSee devs that they made huge mistake in modern versions.
Once they fix it you'll get all the glory, all other image viewers will kneel, maybe even the libjpeg devs will thank you too.

[Xn User]

Then explaine this https://www.mediafire.com/file/iidfvc7r ... s.avi/file this is subtraction result XnView MP of result ACDSee Pro 5.0

[Xn User]

sure, but you should not keep this secret here, you need to tell ACDSee devs that they made huge mistake in modern versions.
Once they fix it you'll get all the glory, all other image viewers will kneel, maybe even the libjpeg devs will thank you too.

I'm sorry , but you still use coders and decoders with rude work methods ? You do compression to MP3 only very very old codecs and nothing not try compress and decompress MP3 modern codecs ? You compress video in veryvery old MPEG4 codecs even if available MPEG4 codecs with better quality ?
Do you understand that this is stupid?
Yes ACDSee Pro 5.0 with ide_acdstd.apl differs from the original and xn result , but in better side differ
And I offer do also , becausee will be better ) I offer do this optional in settings read JPEG format
so quality lost only if use :
subsampling 2x1,1x1,1x1
or
subsampling 1x2,1x1,1x1
if use
subsampling 1x1,1x1,1x1
no such losses were detected

By the way , such an rude decoder use very many where , for example :
viewtopic.php?p=202804#p202804
lists programs with this rude decoder :

XnView Classic
XnView MP
XnConvert
ACDSee 3.0
ACDSee PowerPack 4.0
ACDSee PowerPack 5.0
IrfanView
FastStone Image Viewer

simply because people not want think how to do better

and in ACDSee Free JPEG decoder work much more rude and similar on XnView MP

[XnTriq]

I'm afraid I don't understand what this request is about.
To my knowledge, dithering is used to prevent color banding and to improve (perceived) image quality during reduction of color depth.
If an image with a color depth greater than true color (= 24 bits = 8 bits per channel) is opened in XnView, there's a message:

This picture will be converted to RGB with 8bits per component

I don't know which method is used to reduce the color in this case.

[user0]

I'm afraid I don't understand what this request is about.

he wants image to be smooother at 100% which is OK request

dithering is not the correct term, but this is only part of the problem
he cannot understand that his ACDSee5 produces different output from freshly installed ACDSee2025
(whether it is due to him messing with IDE_ACDStd.apl or ACDSee devs finally fixed it)