Web Analytics

shooting LED wall - banding and sawtooth artefact


Video Assist Hungary
 

Hi

I have worked with LED walls before with Alexa, and had to genlock/TC lock them to get rid of refresh artefacts and double lines. However, now I'm on a mini, and I see horizontal banding with sawtooths on the edges..

the bands move with the camera (so this is sensor-related), change thickness with shutter and backlight, and I cannot get rid of them with genlock or anything else. If the speeds doesnt sync, they keep moving up on the screen


I hope this is the correct place to ask, but this to me seems like an issue between the moving shutter and the duty cycle of the LEDs...and thats a bit over my head.

thank you
Balazs Rozgonyi
Video Assist Hungary


Video Assist Hungary
 

Attaching a picture of the problem:



Balazs Rozgonyi
Video Assist Hungary

On 2019. Oct 16., at 23:41, Video Assist Hungary <balazs@...> wrote:

Hi

I have worked with LED walls before with Alexa, and had to genlock/TC lock them to get rid of refresh artefacts and double lines. However, now I'm on a mini, and I see horizontal banding with sawtooths on the edges..

the bands move with the camera (so this is sensor-related), change thickness with shutter and backlight, and I cannot get rid of them with genlock or anything else. If the speeds doesnt sync, they keep moving up on the screen


I hope this is the correct place to ask, but this to me seems like an issue between the moving shutter and the duty cycle of the LEDs...and thats a bit over my head.

thank you
Balazs Rozgonyi
Video Assist Hungary


dave@...
 

Firstly check the source hertz and frame rate if the source material. I’ve had this when mac a set to 60hertz. Also I remember using a 179.9 degree shutter that was the only angle that Bandubg disappeared. 


hope this helps,
dave 


Video Assist Hungary
 

Thanks Dave

I have full control over source, and it’s 23,976. Player, converter and led controller is genlocked together. If they aren’t, the bands move up and down.

I have played around the shutter and obviously it changes the size of the bands. There are rates when they almost disappear. It’s 153.5 for 23.986. However when moving the camera it’s still very apparent. Also seems to be changing by content.

Balázs Rozgonyi
Video Assist Hungary


Guy Mastrion
 

That’s a head scratcher. Do you have another camera you can point at the wall, to test if it is still apparent? 

If if you have not already done it, maybe a hard reset of the camera will clear it up.

How is your power supply?  
--
Guy Mastrion, DP, Creative Director, Professor


Jonathon Sendall
 

Hey Balazs

Just trying to wrap my head around this as I might be shooting back projected material soon but was thinking of opting for projectors instead of LED walls.

One thing I noticed is that the banding is curved and not absolutley horizontal. One test you might do is to feed a image to the wall, a still not a video, with a hard edge between black and white vertical and then an image with a hard edge black and white horizontal and test again, and then once again a still with some kind of gradient that is high res and doesn't have any banding in the original. I just thought that might eliminate the possiblility of two opposing codecs/frame/sample rates and see if it might be to do with the physical lines of the LED wall interfering with the Alexa sensor read pattern.

Also for the LED wall arent there some seperate controls that are nothing to do with the computer/server that is generating your footage?

Good luck

Jonathon Sendall
DP, London UK.


Video Assist Hungary
 

I have tested with a Mini and an Alexa
We might have cracked it, it appears to be connected to the refresh rate of the wall. The LEDs are driven by a PWM cycle, and you can adjust all the clocks and duty cycles, also the row blanking in microsecs and line changing time. Pushing the refresh rate into the 8000 range helps it considerably. 

Balazs Rozgonyi 
Video assist hungary 


On 2019. Oct 17., at 17:50, Guy Mastrion <guy@...> wrote:

That’s a head scratcher. Do you have another camera you can point at the wall, to test if it is still apparent? 

If if you have not already done it, maybe a hard reset of the camera will clear it up.

How is your power supply?  
--
Guy Mastrion, DP, Creative Director, Professor


Video Assist Hungary
 

One thing I noticed is that the banding is curved and not absolutley horizontal.
The Alexa is a global shutter camera. And these move when you pan the camera up and down, so this is caused by the readout, or more precisely the wall changing during readout, I think. That’s why it’s angled, imho. But I have not seen curves. Some duty cycles in the wall controller seem to change the angle.

Balazs Rozgonyi
Video assist hungary


Art Adams
 

>The Alexa is a global shutter camera.

 

Every Alexa employs a rolling shutter.

 

Art Adams

Cinema Lens Specialist

 

 

ARRI Inc.
3700 Vanowen Street
Burbank, CA 91505
Phone:  (818) 841-7070
Fax:     (818) 848-4028

Get all the latest information from www.arri.com, Facebook, Twitter, Instagram and YouTube.

 

 


This message is confidential. It may also be privileged or otherwise protected by work product immunity or other legal rules. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please send us by fax any message containing deadlines as incoming e-mails are not screened for response deadlines. The integrity and security of this message cannot be guaranteed on the Internet.




This email has been scanned for email related threats and delivered safely by Mimecast.
For more information please visit http://www.mimecast.com


Video Assist Hungary
 

Yes, sorry, that’s what I meant, obviously. Beg pardon. 

Balazs Rozgonyi
Video Assist Hungary. 


On 2019. Oct 17., at 22:50, Art Adams <aadams@...> wrote:



>The Alexa is a global shutter camera.

 

Every Alexa employs a rolling shutter.

_._,_._,_


Art Adams
 

>Yes, sorry, that’s what I meant, obviously. Beg pardon. 

 

Not a problem, just clarifying. 😊 Let us know how you progress. This is an interesting problem. I’m not sure how you’ll get around the camera movement issue.

 

Art Adams

Cinema Lens Specialist

 

 

ARRI Inc.
3700 Vanowen Street
Burbank, CA 91505
Phone:  (818) 841-7070
Fax:     (818) 848-4028

Get all the latest information from www.arri.com, Facebook, Twitter, Instagram and YouTube.

 

 


This message is confidential. It may also be privileged or otherwise protected by work product immunity or other legal rules. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please send us by fax any message containing deadlines as incoming e-mails are not screened for response deadlines. The integrity and security of this message cannot be guaranteed on the Internet.




This email has been scanned for email related threats and delivered safely by Mimecast.
For more information please visit http://www.mimecast.com


Andy Jarosz
 

Just stumbled across this thread, but may have some input. Refresh rate is not actually as important in LED walls as the scan rate is.

LED panels are capable of a much faster refresh than camera shutter speeds, so as long as you’re genlocked, that will be no problem. However, the scan rate of the panel, expressed as a fraction such as 1/12, 1/16, 1/32 etc. shows what percentage of the screen is lit up at any given time. If this number is too low, the shutter of the camera will catch the screen mid-refresh, and you’ll see artifacts. In general, you want a scan rate of 1/12 or better (1/8 is the best I’ve seen, from Silicon-Core.)

This effect is not necessarily a factor for in person viewing, so the vast majority of LED panels don’t optimize it, and many manufacturers don’t even list it in the specs. It’s also a fundamental relationship between the panel, shutter speed, and camera sensor read speed, which means if you’re stuck with that display, you may be out of luck unfortunately.

-- 
Andy Jarosz
MadlyFX Special Effects & Props
Andy@...
708.420.2639
Chicago, IL


 

One thing about LED Walls to watch is morie.

I was filming at an event a few weeks back and the person was standing about 6 foot away and the wall was dancing away. 

FWIW I was on an F55 shooting HD on the long end of a 70 - 200. 

Michael

Michael J Sanders

London based Cinematographer/DP 
Mobile: +44 (0) 7976 269818   
Web: www.mjsanders.co.uk


On 13 Feb 2020, at 17:51, Andy Jarosz <andy@...> wrote:

This effect is not necessarily a factor for in person viewing, so the vast majority of LED panels don’t optimize it, and many


Sean Porter
 

We just wrapped an LED shoot in Thailand where we experienced a sawtooth artifact effect.  we spent many hours troubleshooting and ultimately we believe it was due to the orientation of the panels themselves - they were installed 90deg from their native arrangement and we think the top-to-bottom refresh of the camera was conflicting with the now left-to-right refresh of the display. Panels arranged correctly did not display the sawtooth artifact, so be cognizant of your installations.

hopefully this helps future efforts =)

Sean Porter
DP
Oregon


Video Assist Hungary
 

Sawtooth is a refresh rate issue in the wall. Basically the wall is not refreshing fast (or slow!) enough to fill the camera image. Yes, it’s much more noticeable when the camera is angled to the wall, but even with head on it’s there - only it’s a very thin horizontal line. It’s easier to see on some flat colors - usually greyish or greenish, not very saturated. It’s also visible more when panning the camera up and down. The lines stay stationery on the camera, indicating it is connected to sensor refresh. 
It can be solved by meticulously adjusting the wall parameters and genlock timings with the sensor. 



Balazs Rozgonyi
Fillscrn
Tech director


On 2021. Oct 1., at 20:11, Sean Porter <seancporter@...> wrote:

We just wrapped an LED shoot in Thailand where we experienced a sawtooth artifact effect.  we spent many hours troubleshooting and ultimately we believe it was due to the orientation of the panels themselves - they were installed 90deg from their native arrangement and we think the top-to-bottom refresh of the camera was conflicting with the now left-to-right refresh of the display. Panels arranged correctly did not display the sawtooth artifact, so be cognizant of your installations.

hopefully this helps future efforts =)

Sean Porter
DP
Oregon


Jason Norman
 

Was there a good reason to install them in a different orientation?
Do you recall the led manufacturer?
Who teched your wall and who was responsible for the content?

LED Walls should have a Hz rate locked in step with your camera. At least that’s how we do it on the tv show I work on

Jason M. Norman
I.A.T.S.E 728
Lighting Console Programmer
Network Systems Specialist
Lucid Lighting Co.
Los Angeles, CA
(c) 360.481.5379


Sean Porter
 

Hey James,

Was there a good reason to install them in a different orientation?

Yes and no.  We needed a vertical curved cyc type installation, and the panels were originally arranged in a long horizontal curve.  So to simplify configuration it was decided to split the horizontal wall in half, rotate each section 90 degrees and build the wall that way...  in hindsight, our downfall.

Do you recall the led manufacturer?

They were ROE Diamond panels with a Brompton processor.
 
Who teched your wall and who was responsible for the content?

I'm not going to throw anyone under the bus.  Content was captured via a Meta Two 360 camera, stitched with Mistika and mapped/positioned in Unreal.
 
LED Walls should have a Hz rate locked in step with your camera. At least that’s how we do it on the tv show I work on

We found no difference ultimately whether the camera was genlocked or not.  The panels, processor, etc were all genlocked together running at 48hz.  Camera was at 24fps and we tested multiple shutter angles, ultimately working at 144.

We had another, flat, natively oriented panel array adjacent to the curved wall, and it exhibited no sawtooth artifacts.

At one point my AC cleverly suggested we try the camera on it's side, rotated 90deg (we were using a square portion of the sensor).  This actually got rid of the sawtooth artifacts but introduced other artifacts we didn't have time to address.  In the end we mitigated the effect as best we could in camera and it'll be a combo of DI tools and VFX to clean up anything remaining in the edit.

Cheers
Sean Porter
DP
Oregon