Film-Tech Forum ARCHIVE: Animated Test Files for NEC

This is topic Animated Test Files for NEC in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.

To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=003734

Posted by Justin Hamaker (Member # 2165) on 07-07-2019, 06:13 PM:

I am trying to make an animated checkerboard test file because I have a stuck red pixel on one of my NEC NC2000C projectors. I saw a suggestion in another thread that playing such a clip for an extended period might resolve the issue. I know I can upload PNG files with the TI software, but I was wondering if there is anyway to upload an animated file. Or would I need to create a DCP and play it from the GDC server.

Posted by Marcel Birgelen (Member # 6801) on 07-08-2019, 01:59 AM:

AFAIK the TI engine doesn't really support animated files, so the route should be via a DCP or alternate content input. You can find some of those "DLP repair" videos with some highly animated content on YouTube. (Warning: May not be suited for epileptics.)

I'm probably not supposed to show you how you can download a movie from YouTube, but I've heard it's possible. [Wink]

From there, you can use the known route of creating a DCP from it (DCP-o-Matic). Alternatively, you could switch your input to HDMI and connect your notebook to it, that would save you the process of downloading (possibly illegal) and creating a DCP from it.

I guess it doesn't have to be a checkerboard, it can be anything that gets those mirrors a good workout...

I'm in the realm of the skeptics here, I think the chances of it actually working are pretty slim, but we've heard some sporadic success stories. Let us know if it worked for you.

Posted by Justin Hamaker (Member # 2165) on 07-08-2019, 02:30 AM:

I created a DCP with a pattern I made in Photoshop. I ran it for 10 minutes and it didn't do anything, but I've seen the suggestion to leave it running over night.

As for the YouTube videos, I'm already ahead of you. I used to use this as a source for trailers for a lobby trailer reel. Unfortunately it was a moving target finding the most up to date information on how to download.

Posted by Marcel Birgelen (Member # 6801) on 07-08-2019, 05:48 AM:

Downloading videos from YouTube doesn't help Google's business model, so that's why they're constantly changing their stuff to frustrate downloading the content.

Yes, the idea is that you leave this running for a long time to increase your odds, but you should be able switch off the lamp.

I've fixed some image retention problems on some LCDs in a similar way, but fixing a stuck mirror in a DMD device this way is pretty much a lucky shot.

Posted by Carsten Kurz (Member # 5396) on 07-08-2019, 06:05 AM:

One would guess that a simple (max allowed frame rate) frame flicker between black and full red is the most effective pattern in this case. Also playing it in 3D could help (flashing increases effective pattern change frequency). But I think the time span you play this is more relevant than the actual pattern. Maybe an alternating checkerboard on the pixel level (one pixel on, one off, then flip) could also help, since it also moves individual neighboring mirrors against each other.

- Carsten

Posted by Marcel Birgelen (Member # 6801) on 07-08-2019, 06:18 AM:

quote: Carsten Kurz
Maybe an alternating checkerboard on the pixel level (one pixel on, one off, then flip) could also help, since it also moves individual neighboring mirrors against each other.

It will be interesting to see how that will work out with JPEG2000, not only during compression, but also during decompression on the server.

Posted by Carsten Kurz (Member # 5396) on 07-08-2019, 06:39 AM:

I have done a lot of 2k and 4k per-pixel checkerboard tests a while ago with standard J2K data rates. At least visually, they don't suffer much from compression.

http://www.film-tech.com/cgi-bin/ubb/f16/t002306/p1.html

Unfortunately, most of the sample images that were linked in this thread are invalid links now, but most of those I linked to are still there:

https://www.dropbox.com/s/hpnoer0yirohq5h/4kchecker_j2c_crop.jpg?dl=0
(this is not a 'screenshot', but a decompressed J2C of the test pattern, so you notice that J2C deals with it gracefully. Haven't tested monochromatic patterns, though.)

https://www.dropbox.com/s/gnzhg88cbcp0qhr/4k_lines_sony515.jpg?dl=0

- Carsten

Posted by Marcel Birgelen (Member # 6801) on 07-08-2019, 12:08 PM:

I guess those patterns compress sufficiently not to come close to the edge of DCI bandwidth constrains.

I remember that we've talked about some strange behavior on Doremi hardware in the past, where it was clear that the image that was rendered wasn't the full 4K JPEG2000 image in certain cases.

Posted by Carsten Kurz (Member # 5396) on 07-08-2019, 06:41 PM:

Yes, there was a 4k glitch in the Doremi IMB. Don't know if they ironed it out since.

I think it did never occur with real world footage, though.

- Carsten

Posted by Dave Macaulay (Member # 813) on 07-12-2019, 05:20 PM:

The alternating checkerboard is not a normal pattern, it's internal in the ICP, not part of a Barco clone, and not visible in the pattern folder in the ICP. I don't know, but maybe it's actually in there accessible via the ICP/Enigma control program?