Reported Still problems with resize filters when scrolling (4.38 fine, 4.50+ no)

[Ryoma-Echizen]

Im glad they implemented horizontal/vertical centering, specially good when reading "A4" like manga and comics on a 16:9+ screen but there's this bug still lurking on many iview updates.

4.38 32bit vs 4.54 64bit (everything setup the same)
-Resize options (lanczos)
-use fast resample for image shrinking

*fullscreen options
-fit to screen width
-use resample function for 1st display
-backgrounds transparent
-hide cursor

4.38 32bit
I open the image, go fullscreen mode with fit to screen width activated, the resample works as intended, I can scroll and the program feels fast doing the task.

4.54 64bit (the 32bit version is way slower doing the same things}, like running on a 400Mhz pentium dunno why.
" " " " ", the resample only works as long as I don't scroll, the moment I do tbat the program deactivates the filter until I jump to next image, rinse and repeat.

1-If I uncheck "use resample function for 1st display..." and check resample for zooming, the filter only activates when I scroll, on 1st display the image shows as a MS Paint resize with 0 filters.

2-One thing I notice is that "resample for zooming" really eats processing in somw way as I constantly see the mouse cursor doing the loading animation each time I scroll 1 step (I move the mouse so I don't hide itself).

3-It only works as good as on the 4.38 32bit when checking both resample options for fullscreen mode but much much slower, specially for more complex or bigger images (2nd image).

[Bhikkhu Pesala]

I cannot reproduce the problem with either image. My hardware is nearly 7 years old.

Specs: AMD A10-6800K, 8 Gb (WinSPRLevel= 5.9) on Win 10 64-bit 1909 build 18363.657

[Ryoma-Echizen]

I'm using Windows 7 SP1, Athlon 64 620 2.6Ghz+mobo IGP , 4GB of ram, plenty of resources free. I made similar threads all the way back to iv4.40 IIRC (dunno if you can access my old threads), that's why I never updated my 4.38 version. I'll attach a video of the issue.

Many other images viewers have resize built in with no extra optiosn and all of them work fine and as intended, it should be an easy task for any cpu or even 10+ yo IGP.

The resample on zooming behave the same on both of my iv versions (guess that's fine). So the main issue is that resample is automatically disabled on any scroll command unless both resamples are chosen which are not optimal. OFC, displaying images 1:1 on fullscreen have no issues because there's no resample filter involved. Even when I had to reinstall the OS long time ago I retested this, same result.

This is the video

https://drive.google.com/file/d/1cPlqd7OEeLrF6TSrCVCwzhlfh-Nn2LSx/view?usp=sharing

See the "loading" wheel? resample for zooming always do extraprocessing on top of resampling (never used this in my "functional" iv's). Why resample autodisables itself when I scroll either by wheel or kb when using resample of 1st display (if off, no resample will be used and images will look jaggied, that's fine, unticking that option technically disables any resample filter).


So something messed up between 4.38 and 4.40+.

[Bhikkhu Pesala]

Windows 7 is no longer supported by MS. I cannot reproduce the issue on Windows 10. All I can suggest is sending a report to Irfan Skiljan (See help, About IrfanView), and tell us what he says. Then I can update the thread.

[Ryoma-Echizen]

I'll try that, I discovered a similar behavior in maximized windowed mode fit to screen width, resample for fitting, I can more or less pinpoint that when scrolling in this way (or fullscreen with both resample options enabled) the original image get's into a loop of resample each time a scroll step is applied hence why I see the "loading" animation in the mouse cursor either using the wheel or the right bar.

The Win7 support applies basically to security updates, after 2014, there's no need to update win7, is as stable as a rock. Many people avoid win10 because it's a russian roulette where it can bork itself or introduce more issues than fixes at any new "update".


Is it possible to use a virtual machine to test it?

Thx for all.