CyberLink Community Forum
where the experts meet
| Advanced Search >
Image corruption when use Layer module.
Reply to this topic
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Hi Pix I think that monitor calibration have nothing to do with this problem. As you can see from the sample picture above snake get so much corrupted and smooth gradient in the shadow is totally lost. Just dark areas below some level are cut off.

I reviewed your screen recording but it seems to be very pixelated from youtube compression and fine detail in dark areas and color gradient is lost from the start.
But most of the time you look at the bright side of the image and did not compare well with adjustment tab. Let's say open dark image, look at dark areas with some details in it then do not move but just switch to layers module. Then again switch to adjustment module.

It could be hardware related but I tried on two separate windows installations with ATI(AMD) adapter and with my integrated Intel graphic adapter. Same to Optodata and but he tried with his Nvidia and Intel adapters.
So to me it looks much more like a software problem. Could you please advise what is your Windows version and PhotoDirector version?

I will do one screen recording with snake picture above to see how easy it is to notice the issue. And it stays with export not just in the preview!

This message was edited 2 times. Last update was at Aug 09. 2019 07:27

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Uploaded to Drive, file is named Snake.mp4

https://drive.google.com/open?id=1eVr9kpexepnKFIw9G27vIRQFE29_vRn1
Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Created new screen recording with real night shots demonstrating how destructive to fine details this issue is.
File is uploaded to drive and named: 20190809053957.mp4

https://drive.google.com/open?id=1eVr9kpexepnKFIw9G27vIRQFE29_vRn1
Reply
QC2.0 [Avatar]
Contributor Private Message Joined: Apr 27, 2016 04:02 Messages: 347 Offline
[Post New]
I just tried it on my Windows 10 and Intel graphic setup.

I was dying to find if there were any clear difference between the previews.
Unfortunately, I can't see the differences in adjustment and layers preview, and it seemed my eyes got illusions after staring at those previews for a long time.

I fully understand some pretty professional photographer might need extremely perfect and bit-by-bit precision in color for their image production, but you suggested us decrease the exposure to a really low level to observe if any visual differences you found in such dark area is not practical for a photographing work.

Shooting and editing photos are works require expertise and experiences. However, a photo is for watch, recording life and enjoy.
Do you use PhotoDirector for medical purpose or scentific research?

Live with it will save you from those annoying knots.

May you feel no offense in this reply.
Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Sorry but you really don't get my point here! And there is no offense here!

First I get into the issue when I saw heavy loss in the detail in DARK AREAS with regular night shots as you can see it from my last recording. Just one party continued till early in the morning. Nothing scientific, medical or ultra professional! As you can see on the second post with Snake picture few posts before. First post is with original picture second is with messed colors. And there is absolutely large loss of detail that makes pictures useless. Details that you can see in the JPG are just black in Layers. And also every fine gradient in shadows or in colors like dark sky for example.

Second I provided advice to lower the exposure just in case you do not have dark images on hand. So you just get bright image and lower exposure. Simple as that! Nothing extreme or super ultra mega adjustments applied. Because defect shows up only in DARK AREAS of the image. Usual shadows but not only, could be areas with low light. Also I do not think that -1 to -2 is extreme exposure adjustment but anyway.

I have a lot of examples uploaded that clearly show the issue and optodata also demonstrated it with simple gray gradient
And all this defects are not normal and not observed in any other software that I use.

Please if you like to cooperate take a look at uploaded screen recordings.
And if you and PIX do not have this issue on same version of PhotoDirector then we can consider something else. Like Windows updates and etc.

PS: Also please advise what is your PhotoDirector version.

This message was edited 2 times. Last update was at Aug 09. 2019 05:58

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Today received reply from support and this issue is escalated to engineering team. Seems like they also could not reproduce issue.


Anyone else with similar problem?
Reply
optodata
Senior Contributor Private Message Location: California, USA Joined: Sep 16, 2011 16:04 Messages: 3608 Online
[Post New]
I think we've gotten kind of stuck here, with so many examples and long videos showing many different forms that the artifacts can show up, alongside more videos and other images that don't seem to show any problems.

Also, I was mistaken in thinking that my 7 stripe images was going to show the same problems on all other machines. I still want to try and find a single, clear, reproducable way of showing the issue, but it's challenging to find a way to present it so that the effect is obvious.

The best examples so far usually need to have exposure settings tweaked or posterization applied, but I want to avoid adjustments like those because they visually alter the rest of the image and can make it seem like "normal" images aren't really affected.

One test that's repeatable on 2 PCs here (my desktop and a Surface Pro 3) is to use a grayscale gradient, this one with more shades than my original 7-stripe version.

Open the attached "Grayscale test gradient image" in any photo editor (I've used IrfanView, MS Photos, MS Paint, Paint.net, Paint 3D, Snip & Sketch, PhD7) and you should see a gradient with evenly spaced vertical lines. In PHD 10's Library, Adjustment and Edit screens, the image looks exactly the same as in the other apps.

However, when viewed in the Layers window and without making any adjustments, the spacing between vertical lines is not linear. The lines are far apart at the dark edges but they become bunched up closely together about 1/3 of the way from the edge toward the center. Here's a close up of the left edge in each view:



What this shows is that for some reason - at least on my 2 systems- the luminance scale in the Layers module is not the same as in the other PhD 10 modules (or that other apps use). This is very much like the Full/Limited luminance issue in video that I linked to in my first reply, and this is where CL's engineers should focus their attention.

Can someone else confirm that the vertical line spacing changes when switching between Lib/Adj/Edit and Layers in PhD 10? Also please confirm that the spacing looks equal in any other photo app, and it's only different in PhD 10 Layers.
[Thumb - section-a-mp4-media-player-classic-home-cinema-v1-4-2668.png]
 Filename
section-a-mp4-media-player-classic-home-cinema-v1-4-2668.png
[Disk]
 Description
Grayscale test gradient image
 Filesize
2061 Kbytes
 Downloaded:
4 time(s)
[Thumb - Gradient view.png]
 Filename
Gradient view.png
[Disk]
 Description
Full size screen shot comparison
 Filesize
563 Kbytes
 Downloaded:
4 time(s)

youtube/optodata

DS365 | Win10 Pro | 8 Core/16T i9-9900K (4.8GHz) | RTX 2070 | 32GB RAM | 6TB SSDs

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | HF S200 (HD 30p) | Yi Action+ 4K | 360Fly 4K 360°
Reply
optodata
Senior Contributor Private Message Location: California, USA Joined: Sep 16, 2011 16:04 Messages: 3608 Online
[Post New]
I don't think I can make the issue any clearer than with this screenshot of the attached image:



In the Layers view, the "2%" bar has become the background shade of gray, and the subtle checkerboard visible in the Edit view is also completely missing. I'm very interested to see if any systems display this image identically in the Lib/Adj/Edit vs the Layers view in PhD 10.
[Thumb - Brightness_Elevated.png]
 Filename
Brightness_Elevated.png
[Disk]
 Description
definitive test image
 Filesize
70 Kbytes
 Downloaded:
3 time(s)

youtube/optodata

DS365 | Win10 Pro | 8 Core/16T i9-9900K (4.8GHz) | RTX 2070 | 32GB RAM | 6TB SSDs

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | HF S200 (HD 30p) | Yi Action+ 4K | 360Fly 4K 360°
Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Same to me.

Just reviewed some old night shots without any adjustments and almost every picture look bad in layers module. Details in the sky and in the shadows get corrupted even whole parts of bildings get blacked out. So there is no problem with adjustments at all.

[Thumb - section-a-mp4-media-player-classic-home-cinema-v1-4-2668.jpg]
 Filename
section-a-mp4-media-player-classic-home-cinema-v1-4-2668.jpg
[Disk]
 Description
 Filesize
158 Kbytes
 Downloaded:
1 time(s)

This message was edited 1 time. Last update was at Aug 10. 2019 00:10

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Optodata of course you can show it better! Just add color:

[Thumb - color_test_1.jpg]
 Filename
color_test_1.jpg
[Disk]
 Description
 Filesize
550 Kbytes
 Downloaded:
1 time(s)

This message was edited 1 time. Last update was at Aug 10. 2019 00:39

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
One more.

[Thumb - color_test_2.jpg]
 Filename
color_test_2.jpg
[Disk]
 Description
 Filesize
67 Kbytes
 Downloaded:
1 time(s)

This message was edited 1 time. Last update was at Aug 10. 2019 00:40

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
And one more.

[Thumb - color_test_3.jpg]
 Filename
color_test_3.jpg
[Disk]
 Description
 Filesize
96 Kbytes
 Downloaded:
1 time(s)

This message was edited 1 time. Last update was at Aug 10. 2017 16:55

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
And one more:

[Thumb - color_test_4.jpg]
 Filename
color_test_4.jpg
[Disk]
 Description
 Filesize
337 Kbytes
 Downloaded:
0 time(s)

This message was edited 1 time. Last update was at Dec 28. 2017 17:55

Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
You can find all gradients to test them from google drive:

https://drive.google.com/open?id=1eVr9kpexepnKFIw9G27vIRQFE29_vRn1

In folder Gradients for test.
Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
Two more.




[Thumb - projector_light2-export.jpg]
 Filename
projector_light2-export.jpg
[Disk]
 Description
 Filesize
72 Kbytes
 Downloaded:
1 time(s)
[Thumb - flash-light-flare-theme-in-black-background.jpg]
 Filename
flash-light-flare-theme-in-black-background.jpg
[Disk]
 Description
 Filesize
150 Kbytes
 Downloaded:
1 time(s)

This message was edited 1 time. Last update was at Aug 28. 2017 21:00

Reply
[Post New]
Hello again,

I think we've gotten kind of stuck here, with so many examples and long videos showing many different forms that the artifacts can show up, alongside more videos and other images that don't seem to show any problems.


I agree. We should be trying to narrow the issue. I'm sorry if my input conributed to any confusion.

Here are screenshots of optodata's most recent gradient image. In PhD10, zoom was set to 100% in Library, Adjustment & Layers. These screenshots are 100% crops of the top left corner of the image. Monitor set to 1920x1080.



As stated previously, the difference is not so stark even though it's there. THEN it hit me like a bolt of lightning!

All my previous investigations, including the screen capture, have been using the perpetual version of PhD10 on reasonably capable hardware.

When I tested the same image in PhD365 (on lesser hardware), I got results similar to both your reported observations. Here are the same screenshots taken using PhD365:



I don't believe the issue is hardware related, knowing that optodata's PC is more than capable (!) & that he's also using PhD365.

I'll pass this on to the team at CL (it's possibly why they weren't able to replicate the issue).

PIX PhD CDR ADR DZ MVL PP Forum Moderator
PIX YouTube channel
Reply
optodata
Senior Contributor Private Message Location: California, USA Joined: Sep 16, 2011 16:04 Messages: 3608 Online
[Post New]
Quote As stated previously, the difference is not so stark even though it's there. THEN it hit me like a bolt of lightning!

All my previous investigations, including the screen capture, have been using the perpetual version of PhD10 on reasonably capable hardware.

When I tested the same image in PhD365 (on lesser hardware), I got results similar to both your reported observations.
...
I'll pass this on to the team at CL (it's possibly why they weren't able to replicate the issue).

PIX

That's brilliant! I'm so happy you thought to check on both versions of the program, and from now on there is no doubt that we will need to clarify which licensing version of every CL app people have, in addition to which software version.

PIX, is the perpetual license v10.0.2509? If so, then it seems very likely that the issue appeared in the 10.6(.3010) upgrade for subscribers. Unfortunately, this would just be the latest example of a long standing "tradition" in CL software of randomly breaking perfectly functioning modules every almost single time new features are added. This is so frustrating for us end users! yell
youtube/optodata

DS365 | Win10 Pro | 8 Core/16T i9-9900K (4.8GHz) | RTX 2070 | 32GB RAM | 6TB SSDs

Canon Vixia GX10 (4K 60p) | HF G30 (HD 60p) | HF S200 (HD 30p) | Yi Action+ 4K | 360Fly 4K 360°
Reply
PPCyber [Avatar]
Newbie Private Message Joined: Jan 28, 2019 03:31 Messages: 27 Offline
[Post New]
When I opened the ticket there are fields with exact version that need to be filled. Also there is a screenshot with my PD version. I could not even imagine that someone from support will try to replicate issue on previous or different version ……

Also hardware issue does not make sense since my self and optodata have tested it on very different systems.


Optodata my sad observations about tradition to test beta products over customers seems to be industrial standard but not reserved mark for CL. And this is from simplest free phone apps up to the most expensive enterprise products. Microsoft and Apple are exemplary for this kind of issues. And also in industries such as automotive and many others. Of course this is not an excuse for CL.

We just need to have more tickets opened so they will put it on priority.

This message was edited 3 times. Last update was at Aug 11. 2019 14:26

Reply
[Post New]
PIX, is the perpetual license v10.0.2509? If so, then it seems very likely that the issue appeared in the 10.6(.3010) upgrade for subscribers.

Exactly right optodata

My perpetual version is, indeed, PhD10 Ultra 10.0.2509.0.

I can confirm that the issue raised by PPCyber was introduced with a recent update of PhD365. I was able to test the gradient image in PhD365 10.0.2107.0 and the same display issue in Layers was not present. That was prior to the March '19 update, which was 10.3.2722 (I think).



I'll bundle up my comparative screenshots and provide them to CL.

It's much easier to find an issue when all the information is in front of you!

PIX

This message was edited 1 time. Last update was at Aug 11. 2019 15:57

PhD CDR ADR DZ MVL PP Forum Moderator
PIX YouTube channel
Reply
[Post New]
Further investigation has revealed that the Layers display issue was introduced with the 10.3.2722.0 update. With that update, features were added to the Layers module that allowed object alignment & distribution. Whether that's related or not, I have no idea.



PIX PhD CDR ADR DZ MVL PP Forum Moderator
PIX YouTube channel
Reply
[Post New]
An update on this issue:

The QA team at CyberLink has replicated the issue with PhD365 running the most recent builds. RD/CS says that it “should be resolved in an upcoming release”.

So, that's some positive news.

Thank you (all) for raising & investigating the Layers issue. The information provided in this thread has helped greatly.

PIX PhD CDR ADR DZ MVL PP Forum Moderator
PIX YouTube channel
Reply
Reply to this topic
Powered by JForum 2.1.8 © JForum Team