Blur Busters Forums

Chief Blur Buster wrote: ↑

12 Dec 2022, 16:07

If you know the exact pixel structure of the LCD, such as BGR and RGB, then yes.

But remember those Samsung OD-OLEDS use a triangle pixel structure and this technique won't work. Even when using Windows to attempt to detect displays' current pixel structure,

I tried few RGBCMY line combinations for sub-pixel accuracy but I didn't notice any difference even when eye tracking very close to the screen. I'll do more experimenting.

I though of BGR and triangle pixels structures. Patterns would need some adjustment to work on those, but also those are some interesting alternatives you mentioned.

Discorz wrote: ↑

13 Dec 2022, 06:29

I tried few RGBCMY line combinations for sub-pixel accuracy but I didn't notice any difference even when eye tracking very close to the screen. I'll do more experimenting.

For RGB displays, you need:

(RGB)(RGB)(RGB)(RGB) Solid 1px white
(RGB)(RGB)(RGB)(RGB) 1px Cyan line + 1px Red line (1/3 offset right)
(RGB)(RGB)(RGB)(RGB) 1px Blue line + 1px Yellow line (2/3 offset right)

White = RGB(255,255,255)
Cyan = RGB(0,255,255)
Red = RGB(255,0,0)
Blue = RGB(0,0,255)
Yellow = RGB(255,255,0)

You never use green nor magenta to ClearType an edge of a 1+ pixel fullwhite thick line on an RGB display.

Easiest is to chart the pixels like the above, you can clearly see the colors you need. Never blindly experiment -- always use graph paper when pixelarting ClearType. On your graph paper, every 3rd column is red, every (3rd + 1) column is green, and every (3rd + 2) column is blue. Then pixelart accordingly. Use color sharpies or use letters R, G, B, when you pixelart subpixelly.

Now, this can even be extended to more complex work:

For pixelarting complex objects to be converted into a ClearType-style monochrome image full of 1/3rd pixel offsets, you can also use MS Paint, Paint.net, PhotoShop, and create a 2-layers where top layer is blank (either solid white or solid black) and next layer is vertical 1-pixel stripes of red, green, blue. Use a delete-pixel tool on the 2nd layer to reveal the colored pixel underneath. Once you've pixelarted your subpixel art (whether lines or shapes), you can use a bunch of masking tricks and point scaling to merge the layers to create a subpixel-aware image (instructions vary by paint program, this is the most complex step). It's almost like the Apple II half-pixel-offset trick, reinvented, for third-pixel offsets by piggybacking on subpixels.

Be warned, this won't work well on current OLED pixel structures because of Samsung's triangle structure and LG's quadruple-pixel "RWBG" pixel processing. So you've created a test that works well on LCDs but not as well on OLEDs, and got the conundrum that ClearType works on LCDs better than OLEDs (for now...)

Chief Blur Buster wrote: ↑

29 Jan 2023, 19:29

Yes, it is a great database!

I do wish more sync tracks was visible, because the accuracy of the pursuits in the database vary widely. Error margins are measurable if the sync track is visible.

But sometimes you gotta take what you can get, e.g. a reviewer contributing a pursuit photo that doesn't have a sync track visible, and thus unconfirmable camera tracking error margin.

Unfortunately it is what it is.

One of the biggest issues IMO is static camera recording. I'm aware in some cases it can work, but it should be avoided 99% of the time due to human eye vs camera shutter incomparability. On top of this people do stationary recording on landing Compare Frame Rates page. So for me these automatically go down the drain. Now after seeing bunch of ufos I'd dare to say there are more faulty static ufos online than pursuit ufos. Even googling for "ufo motion test" images is enough to see the poor static vs tracking ufo ratio. PLEASE consider a solution for this major issue in next update!

I guess the simplest solution is to add a self explanatory image/animation in the description area because people usually fly past the texts. This gets things straightforward right off the bat, at least for the pursuit part.
A note about optimal camera shutter speed can also be put here, or even note to include sync track/not crop it out. This in combination with my previously suggested compact ufo pattern should do the trick, hopefully.

It would be nice to see a Ghosting Test as main TestUFO page or add sync track to Compare Frame Rates (but note that overshoot artifacts are not accentuated on black patterns). Otherwise there needs to be a note in bold letters to switch to Ghosting Test for proper pursuit camera.

Discorz wrote: ↑

30 Jan 2023, 16:38

Chief Blur Buster wrote: ↑

29 Jan 2023, 19:29

Yes, it is a great database!

I do wish more sync tracks was visible, because the accuracy of the pursuits in the database vary widely. Error margins are measurable if the sync track is visible.

But sometimes you gotta take what you can get, e.g. a reviewer contributing a pursuit photo that doesn't have a sync track visible, and thus unconfirmable camera tracking error margin.

Unfortunately it is what it is.

One of the biggest issues IMO is static camera recording. I'm aware in some cases it can work, but it should be avoided 99% of the time due to human eye vs camera shutter incomparability. On top of this people do stationary recording on landing Compare Frame Rates page. So for me these automatically go down the drain. Now after seeing bunch of ufos I'd dare to say there are more faulty static ufos online than pursuit ufos. Even googling for "ufo motion test" images is enough to see the poor static vs tracking ufo ratio. PLEASE consider a solution for this major issue in next update!

I guess the simplest solution is to add a self explanatory image/animation in the description area because people usually fly past the texts. This gets things straightforward right off the bat, at least for the pursuit part.

testufo-stationary-vs-tracking.png

A note about optimal camera shutter speed can also be put here, or even note to include sync track/not crop it out. This in combination with my previously suggested compact ufo pattern should do the trick, hopefully.

It would be nice to see a Ghosting Test as main TestUFO page or add sync track to Compare Frame Rates (but note that overshoot artifacts are not accentuated on black patterns). Otherwise there needs to be a note in bold letters to switch to Ghosting Test for proper pursuit camera.

Fantastic suggestion. I've added it to my ticket system to use a variation of this image.

TestUFO is currently undergoing a major revision to support HDR, not sure I can slipstream this in (yet) but this is a fantastic suggestion.