View Full Version : Multi monitor enviroment
Stacey Spears
02-26-2008, 03:12 PM
I ran into a problem with REDCINE and multi-mon. I want to ensure that my future plans will work with SCRATCH, which is:
1. Sever 2003 SP2
2. Primary display @ 2560x1600
3. Secondary display via HD-SDI (Quadro FX 4600) @ 1920x1080
Is that a supported scenario?
On a side note, I just got the FX4600 SDI card and the levels over SDI are incorrect. I am feeding the output into a Tek WFM700. I need to speak with nVidia about this. Have you confirmed the levels out of the SCRATCH approved SDI cards?
Adam Glick
02-26-2008, 03:25 PM
Server 2003? Are you trying to run on a 4-way or 8-way system?
I have not heard of problems with the 4600SDI as far as levels are concerned. Is your source footage "safe" to begin with? When's the last time you calibrated the TEK WFM?
Here's something Lucas posted some time back on scratch-list about using a LUT to clip @ broadcast safe levels...
----------------------------------------
\ Hi all,
With the recent thread addressing LUTs as a legalizing method, I have
posted
four files on the SCRATCH List page.
http://groups.google.com/group/scratch-list/files
Clipping1D.lut
Highlightillegal.lut
Clipping1D.xls
Highlightillegal.xls
Clipping1D.lut will put a hard clamp on output at 95 and 685.
Highlightillegal will highlight anything below 95 as pure red, and
anything
above 685 as pure blue. The .XLS files are the Excel files that I
created
the LUTs from. These are not meant to be used as output legalizers,
per se,
but are put there as reference so everybody can see that it is fairly
straightforward to create these files and use them for this purpose.
To use these files, I'd recommend loading the 1D LUTs into the 3D LUT
section in the LUT page so the entire screen doesn't go out of whack -
just
the image. Yes. you can load 1D LUTs into the 3D LUT section, and
SCRATCH
will convert the LUTs on the fly. ;)
Also remember that with the LUT Export Extension, you can create your
own 1D
or 3D LUTs directly out of SCRATCH. Load up appropriate test frames
and
create a grade that ramps or clips or does whatever you think is
necessary
and appropriate for your output signals and then save as a LUT!
Cheers,
Lucas Wilson
Stacey Spears
02-26-2008, 03:46 PM
I am using Server 2003 because I am using direct attached storage. Its a 9.5 TB RAID50. XP can't handle >2 TB in a volume. The config is a dual 2.8 GHz quad core. The MB is the Supermicro X7DWA-N.
I will double check the levels on the WFM700 with a Tek HD signal gen.
Lucas Wilson
02-26-2008, 03:46 PM
I ran into a problem with REDCINE and multi-mon. I want to ensure that my future plans will work with SCRATCH, which is:
1. Sever 2003 SP2
2. Primary display @ 2560x1600
3. Secondary display via HD-SDI (Quadro FX 4600) @ 1920x1080
Is that a supported scenario?
On a side note, I just got the FX4600 SDI card and the levels over SDI are incorrect. I am feeding the output into a Tek WFM700. I need to speak with nVidia about this. Have you confirmed the levels out of the SCRATCH approved SDI cards?
Hey Stacey,
Two issues here:
1) REDCINE is not a multi-mon app. SCRATCH is. You will have no issues w/SCRATCH.
2) Levels on 4600-SDI -- how are you testing this? What color space, what output, what input source, etc, etc.?
Lucas
Joe Carney
02-26-2008, 06:00 PM
Stacey, are you able to install the fx4600 drivers in Server 2003 sp2?
If so, what did you have to do?
Joe C.
p.s. I just purchased a non sdi FX4600.
Stacey Spears
02-26-2008, 07:50 PM
Zeke,
You use the XP drivers on Server 2003.
Lucas,
I can't seem to locate the app to grab stills from the WFM700, but I will explain what I see.
I am feeding a Tek TG700 signal gen into B and the nVidia SDI into A. Both are set to 1080p23.976.
On the TG700, I am sending SMPTE bars. I have chosen line select and grabbed line 1000, which contains the PLUGE and the 100% box. The 100% box (Y 235, Cb/Cr 128 (in 8-bit)) lands on the 100% marker on the WFM700. The black (Y 16, Cb/Cr 128 (in 8-bit)) lands on the 0% marker. You can also see the -4% and +4% fall below and above 0%, were they belong.
On the nVidia SDI, I have chosen DualView since I don't have an app that uses the SDI API directly. I used a BMP of SMPTE bars as my desktop background. The 100% box (RGB 235 (in 8-bit)) lands below 100%. The black (RGB 16 (in 8-bit)) lands above 0%. You can also see the -4% and +4% are both above 0%.
If I were a betting man, they are mapping RGB 0 to 0% instead of RGB 16. They are mapping RGB 255 to 100% instead of RGB 235.
Over the years I have gone round and round with them and they do the correct thing over DVI. Y 16, Cb/Cr 128 is converted to RGB 16 and so on.
They had compressed behavior on the analog VGA output, which I understood, because they were not using a dual rail power.
Perhaps using a native SDI application would place the levels where they belong on the scope, can you confirm? I have attached the bmp, in case you want to bring it into SCRATCH and send the SDI output and check the levels.
Lucas Wilson
02-27-2008, 08:13 AM
Hey Stacey,
Maybe I'm confused, but..
Mapping 0/255 to 16/235 is only valid for NTSC methodology. For anything having to do with datacentric high-resolution workflow, e.g.: DPX or R3D, output should be a one-to-one mapping. In that case, 0 should map to 0, not 16, and 255 should map to 255, not 235.
The SMPTE bars that you attach bear this out. Those are 75% bars, with black at 16 and white at 235. When you look at it on a WFM700 in 75% bars mode, then it will scale to 0 and 100. I think a better gauge of exactly what's going might be tracking IRE. What API matrix are you using to output that to 1080p/23.976? Are you perhaps using the unscaled matrix? Unscaled, I think it *should* be a 1-to-1 mapping and what you are seeing is correct.
Lucas
-----
ASSIMILATE, Inc.
LA, CA, USA
Joe Carney
02-27-2008, 06:43 PM
In case anyone is curious, here is something new from Tyan.
http://www.tyan.com/product_board_spec.aspx?pid=560
double the memory slots, only 1 gigabit connection. PCIe slots can be used for non graphics cards.
Seems you could use the less expensive 1Gig FB-DIMM if necessary.