Image Sharpness Setting Option? Bring Back Moire?
OK, I'll bite. Let's just follow that thought through. If you think I'd get different results from differently calibrated monitors, plugged into the same computer, do you think they're changing the actual RGB values of the pixels that are being displayed on screen? I have two monitors attached to my PC that are slightly different versions of the same model (one a year or so older) and while they have been calibrated to be close, they are not identical. Same results of the changed-by-pano2vr pixel data RGB values on each.Neal wrote: ↑Fri Jul 31, 2020 12:17 am
Calibrating a monitor is more than just changing the brightness and tint values on the monitor, it changes the color profile. The color profile is what maps the RGB values in the file, to what colors are displayed in the browser. The ICC is also part of that change of adjustments. So yes, if the color profile is off, a value of 0,0,0 in the image can be displayed as 15,15,15 in the browser because the color profile instructed the browser to use that value. --adding this for clearification. It is also possible to make 255, 255, 255 look like 155,155,255 by changing the tinting on the monitor without changing the RBG values.
Now, not all graphic programs use the color profile or the ICC date. However, by default Firefox only enables color management for images tagged with ICC data. Edge handles color management differently. It converts everything to sRBG but doesn't use the monitor profile. As for Chrome, it ignores them.
What about different PCs all running the same Firefox browser and looking at the same panorama? Do you think the RGB values would change?
It's honestly really not going to make any difference to the source data that's being displayed.
I think you're combining the separate ideas of what you see with what is being displayed.
Just because I put on sunglasses does not mean that the colours of some roses I look at have actually changed. That's just like changing the brightness value on your monitor. But, two pictures of those roses taken by a camera at different exposure settings and capturing them as data would produce different RGB values for them.
Pano2VR is taking an image with exposure setting A and displaying as if it had exposure setting B by changing the RGB values.
You, you can test this with your own data. You don't even need to do anything complicated.
Create a panoramic image that is 4000x2000 pixels (size doesn't really matter, actually) and make it completely black, RGB 0,0,0.
Now make the panorama tiles and check the RGB values there. They will all be RGB 0,0,0. Everywhere, at every resolution.
Display this panorama and take a screenshot. Check the RGB values with an eye dropper in your image editing software. They will be 15,15,15.
That's it. It's changed. Now why it changes is the next puzzle...
1) Is the GPU changing the pixels passed to it by pano2vr_player.js?
2) Is the GPU not being used and a software method is changing the pixels passed to it by pano2vr_player.js?
3) Is pano2vr_player.js changing the pixels before passing them to the GPU (or software method)?
4) Is the library that pano2vr_player.js calls to pass the pixels to the GPU (or software method) changing them on the way?
5) Is the browser somehow interfering with the library calls from pano2vr_player.js and causing them to be changed as they're passed to the GPU/software?
6) Do different browsers interfere in different ways?
7) Do different platforms have different results or are they all affected in the same way?
Etc. etc. etc.
We can test some of those options out by trying a test panorama that is pure black in different browsers and different platforms and taking screenshots, and screengrabs on PCs, Macs, Tablets, and Phones. You know, like the developers of this program should have done years ago and at regular intervals since; as standards, libraries, browsers and hardware changes.
Then we move on to Problem B...the images are also more blurry than they need to be compared to other viewers and panorama software using the same data at the same resolution on the same monitors and with the tiles created at the optimum 1:1 pixel ratio of the size of the cubes images.
Could it be something as simple as: the size of the canvas that everything is written to in memory is only say, 90% of what is finally displayed, then it is stretched up to 100%, which automatically causes it to be slightly softened as the extra 10% of pixels are interpolated and we're never able to get that sharp 100% 1:1 pixel ratio that we should be getting?
Could be lots of things, but the one thing we can be sure of: PC, Firefox = RGB 0,0,0 Image tiles display as RGB 15,15,15.
Fixing that would allow some wider (and identical to the supplied image) contrast, which would automatically improve the image.
For those who say "It's not very different" how much difference is acceptable? 'Blacks' capped at RGB 30,30,30 and all whites capped at 220,220,220 instead of 255,255,255?
Just how bad do you want your images to look before you ask for it to be fixed?
So...you're agreeing with me? And it doesn't matter if someone is colourblind because Pano2vr is definitely, confirmed to be futzing with the RGB values and people who are not colourblind will see the differences? Cool!360Texas wrote: ↑Thu Jul 30, 2020 11:54 pm Branigan wrote
RGB numbers are objective arithmetic valuesconfirming for you in RGB numbers what my eyes can see.
what my eyes can see are quite subjective.
What your eyes can see might be quite different than what my eyes can see... keeping in mind that about 20 years ago I did have my cataract issues resolved. Post surgery I was able to see colors and sharpness quite vividly.
Looks like I'm going to have to make an animated GIF to show the differences toggling on and off so people can see with whatever quality or quantity of eyes they have, on whichever monitor: it's different. Both colour-wise and blurriness-wise.
- 360Texas
- Moderator
- Posts: 3684
- Joined: Sat Sep 09, 2006 6:06 pm
- Location: Fort Worth, Texas USA
- Contact:
This is what I wrote - full text quotation with added emphasis. All I wrote was that the numbers were the same. I did not agree with your work because of questioning the capture programs apparent issues.
Please revisit Neals comment about capture programs known issues.
Please revisit Neals comment about capture programs known issues.
Interesting observations
Yes I did use a program called "Pixie" and did find that the "your image sample" did result in
Original RGB Black as 0,0,0 White 255 x RGB.
Screen capture "pixie" did produce Black 15,15,15 White 235 RGB
That begs the question... Is this an issue for the screen capture not rendering properly ?
Yes, your screen capture and my capture program seem to produce same values... maybe we should question the screen capture programs capture techniques ? Read as why 15,15,15 and 235 not as original 0,0,0 and 255 ?
Dude, you keep asking me to create a pano or modify a pano and measure the values. I did that, I posted a link to that pano and have been asking you for the last 3 posts to also measure the values. I know what those values were at the source, and I know the values that are displayed in the pano on my computer. There are also fine lines to check the blurriness of the image, gradients to check for banding and curves for smoothness.Branigan wrote: ↑Fri Jul 31, 2020 10:50 amOK, I'll bite. Let's just follow that thought through. If you think I'd get different results from differently calibrated monitors, plugged into the same computer, do you think they're changing the actual RGB values of the pixels that are being displayed on screen? I have two monitors attached to my PC that are slightly different versions of the same model (one a year or so older) and while they have been calibrated to be close, they are not identical. Same results of the changed-by-pano2vr pixel data RGB values on each.Neal wrote: ↑Fri Jul 31, 2020 12:17 am
Calibrating a monitor is more than just changing the brightness and tint values on the monitor, it changes the color profile. The color profile is what maps the RGB values in the file, to what colors are displayed in the browser. The ICC is also part of that change of adjustments. So yes, if the color profile is off, a value of 0,0,0 in the image can be displayed as 15,15,15 in the browser because the color profile instructed the browser to use that value. --adding this for clearification. It is also possible to make 255, 255, 255 look like 155,155,255 by changing the tinting on the monitor without changing the RBG values.
Now, not all graphic programs use the color profile or the ICC date. However, by default Firefox only enables color management for images tagged with ICC data. Edge handles color management differently. It converts everything to sRBG but doesn't use the monitor profile. As for Chrome, it ignores them.
What about different PCs all running the same Firefox browser and looking at the same panorama? Do you think the RGB values would change?
It's honestly really not going to make any difference to the source data that's being displayed.
I think you're combining the separate ideas of what you see with what is being displayed.
Just because I put on sunglasses does not mean that the colours of some roses I look at have actually changed. That's just like changing the brightness value on your monitor. But, two pictures of those roses taken by a camera at different exposure settings and capturing them as data would produce different RGB values for them.
Pano2VR is taking an image with exposure setting A and displaying as if it had exposure setting B by changing the RGB values.
You, you can test this with your own data. You don't even need to do anything complicated.
Create a panoramic image that is 4000x2000 pixels (size doesn't really matter, actually) and make it completely black, RGB 0,0,0.
Now make the panorama tiles and check the RGB values there. They will all be RGB 0,0,0. Everywhere, at every resolution.
Display this panorama and take a screenshot. Check the RGB values with an eye dropper in your image editing software. They will be 15,15,15.
That's it. It's changed. Now why it changes is the next puzzle...
1) Is the GPU changing the pixels passed to it by pano2vr_player.js?
2) Is the GPU not being used and a software method is changing the pixels passed to it by pano2vr_player.js?
3) Is pano2vr_player.js changing the pixels before passing them to the GPU (or software method)?
4) Is the library that pano2vr_player.js calls to pass the pixels to the GPU (or software method) changing them on the way?
5) Is the browser somehow interfering with the library calls from pano2vr_player.js and causing them to be changed as they're passed to the GPU/software?
6) Do different browsers interfere in different ways?
7) Do different platforms have different results or are they all affected in the same way?
Etc. etc. etc.
We can test some of those options out by trying a test panorama that is pure black in different browsers and different platforms and taking screenshots, and screengrabs on PCs, Macs, Tablets, and Phones. You know, like the developers of this program should have done years ago and at regular intervals since; as standards, libraries, browsers and hardware changes.
Then we move on to Problem B...the images are also more blurry than they need to be compared to other viewers and panorama software using the same data at the same resolution on the same monitors and with the tiles created at the optimum 1:1 pixel ratio of the size of the cubes images.
Could it be something as simple as: the size of the canvas that everything is written to in memory is only say, 90% of what is finally displayed, then it is stretched up to 100%, which automatically causes it to be slightly softened as the extra 10% of pixels are interpolated and we're never able to get that sharp 100% 1:1 pixel ratio that we should be getting?
Could be lots of things, but the one thing we can be sure of: PC, Firefox = RGB 0,0,0 Image tiles display as RGB 15,15,15.
Fixing that would allow some wider (and identical to the supplied image) contrast, which would automatically improve the image.
For those who say "It's not very different" how much difference is acceptable? 'Blacks' capped at RGB 30,30,30 and all whites capped at 220,220,220 instead of 255,255,255?
Just how bad do you want your images to look before you ask for it to be fixed?
Here is the link again: https://hvolnew.com/Images/ResolutionTe ... index.html
To be clear, the link takes you to a pano, display through the Pano2VR javascript interface, and was created using Pano2VR.
Get the values for the solid black, white, blue, red, and yellow boxes. Are you seeing banding in the gradient boxes? Are the lines at the bottom sharp or fuzzy?
The RGB values changing are not something I just made up, read the first link I posted above.
Last edited by Neal on Fri Jul 31, 2020 3:53 pm, edited 1 time in total.
- 360Texas
- Moderator
- Posts: 3684
- Joined: Sat Sep 09, 2006 6:06 pm
- Location: Fort Worth, Texas USA
- Contact:
I do not think I was referencing your SVG vector graphic panorama work.
Neal's links here:
https://www.frontify.com/en/blog/are-yo ... lor-codes/
https://www.color-management-guide.com/ ... come%2010
Neal's links here:
https://www.frontify.com/en/blog/are-yo ... lor-codes/
https://www.color-management-guide.com/ ... come%2010
I did a crazy check on myself. (OK, crazy as related to my comments on changing RGB values, no guarantee on anything else).
I went out and recalibrated my monitors. I calibrate using an external device, an i1display sensor. That changed the color profiles for each monitor. Also, If you don't use some type of external device, you are not calibrating your monitor, you are just adjusting it.
The RBG values in the test pano changed as displayed in my browser. In fact, I can move the browser from one monitor to another and see the values changing. This was for Edge.
So I pulled up Firefox, placed in on the same monitor as Edge, and took another reading. Guess what, I got RGB values in Firefox different than I did in Edge.
I did the same test with Chrome, it gave the same values as Edge. (They use the same backend now)
The RGB values were found in each browser using the Colorzilla extension.
I went out and recalibrated my monitors. I calibrate using an external device, an i1display sensor. That changed the color profiles for each monitor. Also, If you don't use some type of external device, you are not calibrating your monitor, you are just adjusting it.
The RBG values in the test pano changed as displayed in my browser. In fact, I can move the browser from one monitor to another and see the values changing. This was for Edge.
So I pulled up Firefox, placed in on the same monitor as Edge, and took another reading. Guess what, I got RGB values in Firefox different than I did in Edge.
I did the same test with Chrome, it gave the same values as Edge. (They use the same backend now)
The RGB values were found in each browser using the Colorzilla extension.
OK, I tried it and on this strip got RGB 0,0,0 on the Blacks, so now....I have no idea what the flip is going on.Neal wrote: ↑Fri Jul 31, 2020 3:21 pm Dude, you keep asking me to create a pano or modify a pano and measure the values. I did that, I posted a link to that pano and have been asking you for the last 3 posts to also measure the values. I know what those values were at the source, and I know the values that are displayed in the pano on my computer. There are also fine lines to check the blurriness of the image, gradients to check for banding and curves for smoothness.
Here is the link again: https://hvolnew.com/Images/ResolutionTe ... index.html
.
Good, that gives us another clue as to what is going on.
Just so you know, the RGB values for each of those colored strips are (255,0,0), (255,255,0), and (0,0,255).
If I used the default windows color profiles, I get values very close to those. (254, 0, 0), (255, 255, 0) and (0,0,254).
If I use the calibrated color profiles, they look more correct but I get different values.
Now, I want to eliminate two more possibilities. What editor did you use to create your black square? Did you set the color by RBG value or by selecting a swatch?
One more question, when you created that square, did you edit the jpg file or create a new one?
Just so you know, the RGB values for each of those colored strips are (255,0,0), (255,255,0), and (0,0,255).
If I used the default windows color profiles, I get values very close to those. (254, 0, 0), (255, 255, 0) and (0,0,254).
If I use the calibrated color profiles, they look more correct but I get different values.
Now, I want to eliminate two more possibilities. What editor did you use to create your black square? Did you set the color by RBG value or by selecting a swatch?
One more question, when you created that square, did you edit the jpg file or create a new one?