[v24] Compare 100% attribute
Moderator: xnview
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
[v24] Compare 100% attribute
Nicely designed image comparison function.
I thought you might add a comparative analysis. If the two were 100% compatible, the app would write “100% identical” or something similar. And if they were incompatible, an icon would appear, and after clicked, a “map” of differences would be displayed.
I thought you might add a comparative analysis. If the two were 100% compatible, the app would write “100% identical” or something similar. And if they were incompatible, an icon would appear, and after clicked, a “map” of differences would be displayed.
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
Please check v25
-
masterjp
- Posts: 665
- Joined: Fri Feb 13, 2009 4:37 pm
- Location: Duesseldorf, Germany
Re: [v24] Compare 100% attribute
The new image comparison analysis yields some somewhat unusual results in isolated cases. I had expected a significantly higher percentage.
PC: Intel 8700k + Asus Z370-F + 48 GB RAM Corsair+G.Skill + Asus RTX 3050 OC
OS: Windows 11 Pro 64bit 25H2 |
GFX: XnViewMP 1.11.5 | Adobe Photoshop Elements 2024 + Elements XXL 11
Android 8-16: Samsung S7, S10, S23, TAB S7, S11, Ugoos AM7 TV-Box
OS: Windows 11 Pro 64bit 25H2 |
GFX: XnViewMP 1.11.5 | Adobe Photoshop Elements 2024 + Elements XXL 11
Android 8-16: Samsung S7, S10, S23, TAB S7, S11, Ugoos AM7 TV-Box
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
Its not like 'similar' in XnView, it's pixels comparemasterjp wrote: Mon Aug 03, 2026 2:47 pm The new image comparison analysis yields some somewhat unusual results in isolated cases. I had expected a significantly higher percentage.
-
masterjp
- Posts: 665
- Joined: Fri Feb 13, 2009 4:37 pm
- Location: Duesseldorf, Germany
Re: [v24] Compare 100% attribute
Okay, I expected a similarity compare from XnViewMP for windows, which is very good.
Pixel compare is also good for the start of xnview mobile.
Pixel compare is also good for the start of xnview mobile.
PC: Intel 8700k + Asus Z370-F + 48 GB RAM Corsair+G.Skill + Asus RTX 3050 OC
OS: Windows 11 Pro 64bit 25H2 |
GFX: XnViewMP 1.11.5 | Adobe Photoshop Elements 2024 + Elements XXL 11
Android 8-16: Samsung S7, S10, S23, TAB S7, S11, Ugoos AM7 TV-Box
OS: Windows 11 Pro 64bit 25H2 |
GFX: XnViewMP 1.11.5 | Adobe Photoshop Elements 2024 + Elements XXL 11
Android 8-16: Samsung S7, S10, S23, TAB S7, S11, Ugoos AM7 TV-Box
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
Re: [v24] Compare 100% attribute
You're making changes so quickly that I suspect you're using an AI model to help you write the code. I don't know which of you two is at fault, but the comparison function isn't working properly.
I converted the jpg photo to lossless png, and then from the png I converted it to lossless jxl and webp. XnMobi shows differences in these files, except comparison with jxl which does not work at all.
When I enlarge each of these files individually, they appear identical (except for jxl, which I cannot enlarge as much as png and webp). When enlarging e.g. webp with png (attachment), the png file is clearly worse than webp.
On XnPC, all three lossless files have the same CRC and look identical when enlarged.
But the original jpg has a different CRC. I converted this jpg to png on PC and both have identical CRC.
It's really nice that you make these programs; I admire and respect you, but it would also be nice if you checked the AI's work properly.
Best regards.
I converted the jpg photo to lossless png, and then from the png I converted it to lossless jxl and webp. XnMobi shows differences in these files, except comparison with jxl which does not work at all.
When I enlarge each of these files individually, they appear identical (except for jxl, which I cannot enlarge as much as png and webp). When enlarging e.g. webp with png (attachment), the png file is clearly worse than webp.
On XnPC, all three lossless files have the same CRC and look identical when enlarged.
But the original jpg has a different CRC. I converted this jpg to png on PC and both have identical CRC.
It's really nice that you make these programs; I admire and respect you, but it would also be nice if you checked the AI's work properly.
Best regards.
You do not have the required permissions to view the files attached to this post.
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
I'm not moving as quickly as I'd like. I use AI assistant for the UI, but for the core/features i work with XnView & my other android tools code...Kadet wrote: Mon Aug 03, 2026 5:45 pm You're making changes so quickly that I suspect you're using an AI model to help you write the code. I don't know which of you two is at fault, but the comparison function isn't working properly.
I've tried, but can't reproduce. Please send me all files to check...I converted the jpg photo to lossless png, and then from the png I converted it to lossless jxl and webp. XnMobi shows differences in these files, except comparison with jxl which does not work at all.
When I enlarge each of these files individually, they appear identical (except for jxl, which I cannot enlarge as much as png and webp). When enlarging e.g. webp with png (attachment), the png file is clearly worse than webp.
On XnPC, all three lossless files have the same CRC and look identical when enlarged.
But the original jpg has a different CRC. I converted this jpg to png on PC and both have identical CRC.
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
Re: [v24] Compare 100% attribute
This same link, like in the Lossless conversion bug:
https://drive.google.com/file/d/1ahUyaS ... sp=sharing
https://drive.google.com/file/d/1ahUyaS ... sp=sharing
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
i've checked on XnView (PC), and the converted files by XnView mobile are fine, there is no visual difference when using 'Compare'.Kadet wrote: Wed Aug 05, 2026 8:09 am This same link, like in the Lossless conversion bug:
https://drive.google.com/file/d/1ahUyaS ... sp=sharing
And on android, compare with original & webp/png, the previous screenshot is with same files???
You do not have the required permissions to view the files attached to this post.
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
Re: [v24] Compare 100% attribute
Yes, the previous screenshot is with the same files. I changed the names on PC for clarity.xnview wrote: Wed Aug 05, 2026 9:09 am And on android, compare with original & webp/png, the previous screenshot is with same files???
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
do you see a difference on PC? and could you post a screenshot between the jpg/png?Kadet wrote: Wed Aug 05, 2026 1:16 pm Yes, the previous screenshot is with the same files. I changed the names on PC for clarity.
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
Re: [v24] Compare 100% attribute
You do not have the required permissions to view the files attached to this post.
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
ok, now i understand. I've added settings for JPEG
Please check v28
Please check v28
-
Kadet
- Posts: 456
- Joined: Thu Oct 20, 2022 7:23 pm
Re: [v24] Compare 100% attribute
Sorry, but the comparison still doesn't work properly. Right now JPG looks just as bad as PNG. So it got worse.
-
xnview
- Author of XnView
- Posts: 48640
- Joined: Mon Oct 13, 2003 7:31 am
- Location: France
Re: [v24] Compare 100% attribute
If you use 'slow int' it will be like previous version for jpeg.Kadet wrote: Thu Aug 06, 2026 7:54 pm Sorry, but the comparison still doesn't work properly. Right now JPG looks just as bad as PNG. So it got worse.
But you need to check on desktop, the jpeg loading on arm uses 'slow int', and i can't change that.
The floating point doesn't give same result on intel & arm. libjpeg recommends using 'slow int'