First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Many sites including LinusTechTips, RTINGS, TomsHardware, and others use the free Blur Busters pursuit camera invention. Now also avaialble as a rail-less smartphone wave, too!
teo
Posts: 20
Joined: 15 Apr 2021, 19:28

First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by teo » 18 Nov 2021, 19:58

alrighty, I need a little guidance on how to better wave my hand.

current tools:
iPhone 8 Plus (1080/60 or 4k/60 video)
A mostly flat box

I understand that I’m looking for full vertical alignment of the sync track. After taking a few screen grabs, I’m realizing that none of “aligned” images are aligned across the entirety of the screen. Which leads me to:
  1. Should I aim to find an aligned image for an individual (i.e. one of top/middle/bottom) ufo? An entire column?
  • If just one, am I better off holding my phone closer to the monitor screen?
  • How do I tell if I’m in focus vs blur?
  • What else stands out that could use improvement from these examples (and how do I improve it)?
These are at 240hz, no strobing, overdrive on “ultra fast.”
240hz ultra fast one
240hz ultra fast one
EE23BF8B-AE2F-4F6E-9D57-E5EE2756022C.jpeg (4.5 MiB) Viewed 839 times
240hz ultra fast two
240hz ultra fast two
1AA6E979-A348-4749-BFA1-1ECE46E54B0A.jpeg (1.46 MiB) Viewed 839 times
240hz ultra fast three
240hz ultra fast three
BEBCC190-69BF-4C63-BBA1-8254EBC6CA63.jpeg (4.96 MiB) Viewed 839 times

User avatar
Chief Blur Buster
Site Admin
Posts: 9829
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Chief Blur Buster » 18 Nov 2021, 20:11

Fantastic first pursuit camera images for a DIY hand-wave smartphone pursuit camera!

You chose the right camera estting, 4K 60fps video (at 1/60sec frame exposure) is a good setting to use for 240Hz pursuit camera via hand-wave iPhone. It produces literally 60 photograph-quality freezeframes per second -- the brute force sample rate increases your likelihood that an imperfect handwave created a near-perfect pursuit photo. (More samples compensating for lack of a camera rail).

They could be better but they are pretty decent attempts.
teo wrote:
18 Nov 2021, 19:58
I understand that I’m looking for full vertical alignment of the sync track. After taking a few screen grabs, I’m realizing that none of “aligned” images are aligned across the entirety of the screen.
Sometimes the tilt of the camera can cause effects that create inconsistencies on the sync track, since not all pixels refresh at the same time (on the display panel, and on the camera sensor). Getting consistency on both is sometimes challenging but what you did is good.

If you switch to a video app, it becomes easier to pre-focus (manual focus) -- iPhone apps like DSLRCamera or ProCam lets you configure manual focus and exposure-per-video-frame.

If you handwave,
- Try holding your phone in portrait mode
- Left hand holding left edge of phone
- Right hand holding right edge of phone
- Start video
- Do not move your arms or shoulders
- Spin your computer chair to follow the UFOs
- Spin back and repeat a few times, about 3 to 5 passes per video clip.
- Jog through the video with your smartphone player's slider and screenshot the clearest frame
(Better to use a player app that can save full-resolution freezeframes, or transfer the video file to the PC and frame-step through it).

Ideally, most of the time, camera lens should be same as human viewing distance (arm's length from lens to screen). However, due to smartphone limitations, and wide-angle issues, you may want to use optical zoom (if you have optical zoom), or move closer to the screen (not recommended for viewing angle sensitive screens like TN).

Some reviewers zoom on just one UFO, while others use an entire column. If you use a portrait smartphone, zoom your camera (or get closer) until top sync track is almost top edge, and bottom sync track is near bottom edge. If you use a landscape smartphone, you may want to zoom into just one row at a time (middle row is the most common). Keep sync tracks visible, they are essentially your certificate of camera tracking accuracy. I can tell you a LOT just by looking at your sync track.

Right now, you need the sync track lines to be perpendicular to the horizontal lines. You're parallelogramming a little bit with vertically straight sync track lines but tilted horizontal lines. So if you're getting best freezeframes from a tilted camera, make suer your sync track lines are perpendicular to the horizontal lines. You can rerorate the whole image in an image viewer if need be, but not essential.

Refer to this images for reference:

Image

So in this case, if you've parallelogrammed the sync track (lines in sync track not 90-degree perpendicular to the horizontal lines), your motion blur error margin is your deviation from the perpenducular red lines. The line disjoint is at the very top/bottom of sync track, so it's harder to see than a mid-line disjoint (e.g. between 2nd and 3rd tickmarks of 4 vertical tickmarks). Your error margin (~25%-ish) is not significant in the case of 240Hz sample-and-hold but becomes a major error margin for strobed pursuit camera photography which can reduce motion blur by about 60% to 90% (depending on selected PureXP level), or up to 99% (on best custom PureXP setting).

Also, if you keep having too much error margin, try reducing camera exposure to only 2 or 3 refresh cycles. You will have gaps in your sync track, but it can then be easier for strobed pursuit camera photography from hand-wave. Multiple refresh cycles are recommended to emulate human vision integration time period for more accurate motion-blur capture. However, for strobed pursuit camera photography, this is far less critical and you can make do with just 2 refresh cycles (1/120sec exposure at 4K 60fps for 240Hz strobe, or 1/60sec exposure at 4K 60fps for 120Hz strobe). Refer to this diagram for error margin of less-than-4-refresh-cycle pursuit camera capture:

Image

In your case, refer to "Missing Mark" for camera exposures shorter than 4 refresh cycles. Technically, the human vision integration time has been in the ballpark (order of magnitude of 1/30sec) originally designed back in the 120Hz days. But getting WYSIWYG photographs of display motion blur remains very accurate for displays at any Hz at pursuit camera capture of 4 refresh cycles (even for temporally dithered displays such as DLP).

So in the event of difficulty doing hand-wave pursuit camera, it's acceptable to shorten camera exposure down to 3 refresh cycles or sometimes even 2, at least when it comes to non-temporally-dithered displays (LCD/OLED). (More refresh cycles per pursuit camera photograph is more important when the display is doing temporals to generate color, so that film can more emulate human vision averaging behaviors).

Quick Tips

- Test out a 3rd party app that lets you control camera exposure per frame (as well as more easily do fixed-focus). Preconfig the camera and start recording before you pursuit a few times per video clip
- Try holding phone with both hands, with a fully stiff body (just spin computer chair to pursuit)
- Use optical zoom where possible instead of coming too close to monitor. You want to simulate eye-to-panel viewing distance, but you can use optical zoom to blow up the UFOs bigger into the resulting photo. However, getting closer to the screen is acceptable if no
- If difficulties persist, please post a cloud link to your video file (GDrive/Dropbox link) or upload to YouTube (so I can single-framestep via "." and "," keys). I will inspect your technique and pull out the best freezeframe, in case it's a freezeframe-selection technique issue rather than a hand-shakiness issue.

Hope this helps!
Once we've selected good pursuit camera attempts, we can crosspost them back to the main XG2431 thread (if you wish)
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

teo
Posts: 20
Joined: 15 Apr 2021, 19:28

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by teo » 18 Nov 2021, 21:43

I appreciate the help. I'll have to re-read the theory a few times before that really sinks in, but for the practical application:
  • I grabbed this $1 "DSLR Camera" iOS app, which does give me helpful sliders for focus and exposure. however, I'm struggling to get it to record in 60 fps which seems to be an important bit. was this the app you had in mind, and have you used this particular app before? it took me a while to find the option for 4k, so maybe I'm missing an 'obvious' setting
  • I'm totally onboard with the chair-spinning technique to get a smoother "wave", but doesn't this now mean I'll only have a chance to have perpendicular sync and horizontal lines within a small arc of that rotation? right now I'm aligning my phone at center screen, so in an unedited frame the left (beginning) and right (end) of a pass will always have some parallelograming
this is the best shot I got from the second round, although I must have either messed up the zoom and/or focus during the recording because it looks both brighter and less clear than my first attempts.
483825DF-447E-4AD8-9392-060F8F3A3D91.jpeg
483825DF-447E-4AD8-9392-060F8F3A3D91.jpeg (3.46 MiB) Viewed 779 times
edit: quick attempt at a 1920 speed which I forgot to put in full screen
non-full screen cropped 1920 speed
non-full screen cropped 1920 speed
2E56737A-ADD8-421E-ACB7-BA41A60839A3.jpeg (340.87 KiB) Viewed 772 times

User avatar
Chief Blur Buster
Site Admin
Posts: 9829
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Chief Blur Buster » 18 Nov 2021, 21:53

teo wrote:
18 Nov 2021, 21:43
however, I'm struggling to get it to record in 60 fps which seems to be an important bit.
You don't *have* to record in 60fps, because exposure-per-frame is more important than the video frame rate.

I admit that I mainly use DSLRCamera at 30fps, or use my professional rail setup (Sony Alpha a6000 mirrorless camera) so there may be a bug in DSLRCamera that prevents reliable 60fps recording, maybe try a trial of a different professional camera app?

We're just merely using video as a stand-in for continuous burst-shooting of photos. We're generating brute-force sheer sample rate (many photos) to compensate for the error margins of shaky hands. High video frame rate means more photo samples per second (like 60 photographs per second). Which can make it easier to find a good pursuit. But it's not critical to use 60fps.

The bonus of 60fps is that it takes less than 30 seconds to generate 1000 pursuit camera photos (in multiple hand waves) in a single video file that is easily shuttled through (playback slider) to find clear pursuits. Doing it at only 30fps just means it takes twice as long to get 1000 photos, so it's only a minor impact to pursuit convenience.

It's the sheer simplicity of embedded camera tracking accuracy information (sync track distortions) in the resulting frames that makes this convenience possible independently of camera frame rate (single photo, burst shoot, or video). The frame rate does not matter except to help you produce more samples to overcome for the inaccuracy of handheld pursuit camera. It's neat you can get sub-millisecond accurate pursuit camera tracking with your hands, simply by way of sheer brute video frame rate. A buffet of freezeframes, of which hopefully one or two or more is a near-perfect pursuit.

30fps is enough too. Heck, 24fps or 15fps video is fine. The important thing is the photo settings per freezeframe, especially exposure. You only have to emulate human vision on a per-frame basis, not on the whole video file.

The human vision integration time is relatively accurately emulated by adjusting camera exposure per frame to capture multiple refresh cycles. Beyond this, video frame rate is only important to capture more photographic samples more quickly in a shorter time period. In other words, how the human vision blends multiple refresh cycles into one blurry soup -- and pursuit camera achieves the WYSIWYG capture of what blurs the human eye saw during eye-tracking (good self-educational animation at www.testufo.com/eyetracking#speed=-1 if you want to stare at bottom UFO for a minute).
teo wrote:
18 Nov 2021, 21:43
center screen, so in an unedited frame the left (beginning) and right (end) of a pass will always have some parallelograming
This is true (horizontally skewed parallelogramming when pursuiting at an angle).

But right now, seeing vertically skewed parallelogramming which is evidence of camera tracking speed mis-sync beyond the parallelogramming caused by oblique angle. I can hand-write what I mean in a sketch, if you get confused.

And make sure vertical lines are perpendicular has similar angle of the top line and bottom line. So if the top line is 95 degrees and bottom line is 95 degrees relative to the vertical line, then you've got a good oblique-angle pursuit. But you're having effects such as 95 degrees to top line and 85 degrees to bottom line, which is a definite camera-tracking error margin independent of oblique angle caused by chair rotation, and is adding a slight amount (the earlier first photos appeared to be very roughly ~+25% more) motion blur versus WYSIWYG. If you are confused, I can sketch.

Sideways parallelogramming (like those in some of your images) = bad
Horizontal keystoning from oblique angle at ends of chair rotation = okay if exactly equally balanced (top and bottom angles identical).
(imagine the "horizontal keystone" mode of a projector, if you've ever used one -- it keeps the topleft/bottomleft corner angles the same, and it keeps topright/bottomright corner angles the same).

There are some minor error margins for oblique-angle pursuits at the extremes of chair spin, but you can get it far less than a 25% motion-blur-error margin if you stick to symmetrical parallelograms (vertical symmetry) like a projector's horizontal keystoning mode, for horizontally oblique angle pursuit photography.

I should sketch this out on paper, and provide it, if you're confused. Chair-spin hand wave pursuits have very little documentation at the moment.

To minimize chair rotation error margin, use more viewing distance (move chair further away from display). That will minimize rotational errors. On the other hand, extra distance may mean lower quality (less resolution per UFO) or noise during zooming. So that's a tradeoff. Generally, I'm OK with camera lens roughly half human viewing distance, when pursuiting IPS panels. This will be a bit more distortion at chair rotation extremes, but can compensate for tiny smartphone lens that don't collect much light per camera exposure. Getting closer allows you to using up more camera sensor surface area per UFO, for better quality photos if you don't have a viewing-angle error margin (e.g. gamma shifts)

Another error margin to keep in mind is slow smartphone sensor scan, which can mean that bottommost sync track may look very different from topmost sync track. But as long as both sync tracks are valid (as per error charts), it's valid. The refresh cycle stacking (integration time) compensates largely to keep it WYSIWYG. The height of the sync track, it's not a meaningful error margin (since the camera sensor scan from top edge of sync track to bottom edge of sync track, is extremely short). As long as all photographed sync tracks above and below your target UFO meets the error margin, the target UFO is accurately recorded.

Nontheless, you captured excellent first-timer pursuit camera photography, all things considered. Most get less accuracy on the first post...
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

User avatar
Chief Blur Buster
Site Admin
Posts: 9829
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Chief Blur Buster » 18 Nov 2021, 22:06

teo wrote:
18 Nov 2021, 21:43
483825DF-447E-4AD8-9392-060F8F3A3D91.jpeg

edit: quick attempt at a 1920 speed which I forgot to put in full screen
2E56737A-ADD8-421E-ACB7-BA41A60839A3.jpeg
Those are much better, albiet slightly overexposed (try using a lower ISO setting). The motion blur error margin looks like less than 10% of a refresh cycle at a quick first glance, perhaps easily under 5%.

You ideally want yellows that don't blow out to nearly white, so when yellows become pale or whiteish, it means camera setting is a bit overexposed. You want the yellows to stay deep yellow. If you're getting grainy images, try increasing screen brightness to compensate for the low ISO setting you hae to use for a specific shutter speed (in manual shutter-priority operation).

Also, zooming sometimes has the disadvantage of adding noise (I noticed), so if you're pushing the limits of your smartphone, it's also acceptable to come a little bit closer to the screen, especially if the screen is sufficiently viewing-angle-insensitive (e.g. IPS instead of TN). So you're playing a game of trade-offs on your camera capabilities for the free $0 rail-less hand-wave smartphone pursuit.

One more round, and then time to try PureXP strobed pursuits?
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

buddha
Posts: 17
Joined: 14 Nov 2021, 15:36

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by buddha » 19 Nov 2021, 07:41

Can you recommend an android app to use? I am struggling with default samsung app.

User avatar
Discorz
VIP Member
Posts: 546
Joined: 06 Sep 2019, 02:39
Location: Europe, Croatia
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Discorz » 19 Nov 2021, 13:41

teo wrote:
18 Nov 2021, 19:58
Great attempt for first time!

Only thing I would suggest is fix camera focus or resolution and reduce ISO or monitor brightness. But some things can still be visible from these pursuits.

I'll make a simplified guide for pursuit camera so I could share it in similar future situations.
>> Compare UFOs << | Monitor Recommendations | Do you use MBR?
GIGABYTE M32Q Review | ASUS VG279QM | DELL AW2518HF | AOC C24G1 | AOC G2790PX | Fujitsu 17P4 CRT | Setup

User avatar
Discorz
VIP Member
Posts: 546
Joined: 06 Sep 2019, 02:39
Location: Europe, Croatia
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Discorz » 19 Nov 2021, 13:45


Simple Guide for Smartphone Hand-Wave Pursuit Camera

To keep it simple; special cameras or equipment are not required. Average smartphone camera with Shutter, ISO and Focus manual adjustments will do the job just fine.
  1. You can either record a video or take a picture of the UFO. Smartphone cameras are usually most flexible and give best results in their 'picture modes'. See what works better and chose one. Make sure camera is set to highest picture quality and resolution. Framerate doesn't matter for video recording.
  2. Shutter speed setting you choose depends on monitor refresh rate, so it needs to be adjusted each time refresh rate is changed. Most importantly set your camera shutter to Hz/4 speed (e.g. for 240Hz s1/60, for 60Hz s1/15...). If camera doesn't have exact 'Hz/4' option then choose one step longer speed (e.g. for 144Hz would be 1/36, choose more common 1/30 instead). Better go longer than shorter.
  3. ISO is simple. Avoid overexposed (too bright, lost detail) and underexposed (too dark, lost detail) image. What you want is something in-between. If you can't find perfect middle better go with underexposed. If image continues to look overexposed reducing monitor brightness will help.
  4. Use Focus setting that gives you most sharp and crisp image. For better image bring your camera closer to screen until all 3 ufos fit in frame like in examples from earlier.
  5. Open Test UFO.
    • Now follow/track/capture the UFO by moving your camera at the same speed. Try to keep your hands steady and consistent. Repeat this multiple times until you're satisfied with the result.
    • Sync Track must be aligned vertically like this or this, avoid misalignment like this or this because the alien will not look accurate or representable to what your eyes see in real life, which is main point of all this. You want the whole line to be vertically flat. End result should look something like this: video or picture. I recommend capturing all 3 ufo rows like in the 'picture' example. But you can do only middle one too.
  6. Most people usually do only one pursuit at maximum Hz, but that is only partial covering of monitor's refresh range. In most cases one overdrive level that is tuned to look great at 240Hz can look terrible at 60Hz or vice versa. For that reason it is best to cover monitor's full refresh range at preferably all or couple of best overdrive levels. Do all these in steps of 60, 85, 100, 120, 144, 165, 180, 200, 240, 300, 360 Hz... (if available). Choose few.

    Fully 'reviewed' monitor would require these pursuits at multiple refresh rates and overdrives combined:
    • VRR enabled (if available) - Note! TestUFO works only with fixed refresh rates, meaning it doesn't support VRR yet, so on some monitors overdrive can behave different in game than in test. If that is the case note it.
    • VRR disabled
    • Blur reduction enabled (if available)
    • Blur reduction and VRR enabled (if available)
    With Blur reduction enabled it is necessary to capture additional top and bottom parts of the screen due to strobe crosstalk. Same goes for Blur reduction + VRR (if available). Majority of monitors have badly tuned other than middle parts of the screen. You can do this in one take like this for video recording. Use fullscreen Strobe Crosstalk Test or Ghosting test but in fullscreen mode.

    But if you want to cut on time and extra work doing only couple is fine.
  7. Footage can be uploaded to website of your choice. If you decided go with video pursuit note that uploading it to sites like YouTube compresses the footage. This is another reason why takin a photo is better. Uncompressed originals are always better.

    When publishing final footage it is important to state monitor and test settings such as:
    • Refresh rate/frame rate
    • Overdrive
    • Screen location, top-middle-bottom (only for blur reduction)
    • UFO speed

This post will be edited!
Last edited by Discorz on 25 Nov 2021, 18:11, edited 4 times in total.
>> Compare UFOs << | Monitor Recommendations | Do you use MBR?
GIGABYTE M32Q Review | ASUS VG279QM | DELL AW2518HF | AOC C24G1 | AOC G2790PX | Fujitsu 17P4 CRT | Setup

User avatar
Chief Blur Buster
Site Admin
Posts: 9829
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by Chief Blur Buster » 19 Nov 2021, 15:25

Until TestUFO supports VRR, an app called Frog Pursuit can be used. It’s VRR compatible.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

teo
Posts: 20
Joined: 15 Apr 2021, 19:28

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Post by teo » 19 Nov 2021, 22:40

after round three I'm at least a little more confident in my camera and monitor settings! I'm still a little undecided on camera distance, but I think somewhere between ~half of normal viewing distance with some zoom and closer-still-with-no-zoom works out a little better than normal distance and high zoom.

I definitely had an incorrect concept of frame rate vs exposure time, so thanks for the extra clarification on 30 vs 60 fps. the manual focus, explicit shutter speed, and ISO slider of the DSLR Camera app seems pretty mandatory to get a better looking image, though the 'half' sample rate (because it only seems to do 30 fps) makes me feel like the search for a good shot is much harder.

I'm getting better at identifying visually when the panning speed is slightly mismatched, but I'm finding it very challenging to adjust my sync line expectations if the parallel white lines are a few degrees off the x-axis. because it's a narrow window in my arc of when those are straight (plus whatever camera shakiness) I often find myself going back and forth between ~3 frames trying to decide which has the sync track the most perpendicular.

I've been using an iOS app called Frame Grabber which is pretty convenient. you do have to specify the export to be at full quality (rather than the default 90%) to avoid compression.

I took a lot of samples, and many of them were terrible. I uploaded two videos plus a dozen or so frame images (which aren't only from those videos)--I'm still figuring out if they're better or not than my previous ones. if anyone would like to peruse the highlights you can find them here.

this one frame caught my eye on my phone because it seemed like the sync track was vertically stacked and mostly perpendicular but the four elements of the sync track seemed more separated. now that I view it on a monitor I'm more inclined to think the pursuit speed is off, and maybe there was some vertical camera movement.
no_zoom.JPEG
no_zoom.JPEG (992.77 KiB) Viewed 410 times

Post Reply