Scopes
Let me start by saying that I am stupid. Well, that might be a bit harsh...DUMB. I made a stupid DUMB mistake. ASSUMPTION rather. And it all had to do with the internal waveform/vectorscopes in FCP.
They aren't accurate. This shouldn't come as a shock to many, and it really wasn't a shock to me...well, OK, it was a little. I thought that I could get away with using the built in scopes on FCP. After all, they looked like they were registering things properly. I wasn't having too many problems with the "blind spots" that Larry Jordan pointed out in one of his monthly newsletters. They looked good...and my picture looked good. And moreover, I did a test output to DVCPRO HD for QC and when they contacted me and informed me of all the hot spots on my footage, I located the spots on the timeline and sure enough, they were over 100IRE. So I adjusted by dropping a Broadcast Safe filter on and chopping off the top end. So I felt confident that all was good.
But then came today...the day of the final mix (to be blogged about later) and layback (process to be touched upon when successful). I was at a post facility using one of their bays and outputting to HDCAM...a bay with proper HD scopes. I got the audio all laid in and we started the output and I was monitoring the footage and..uh oh, the blacks were crushed. VERY low...below 0. Mainly in the blue areas, but often in the red (HD scopes monitor the levels of primary colors separately)...and there were lots of times the REDS and BLUES shot well above 100IRE. But, how was this possible? On my scopes they read fine, and these weren't areas that QC kicked back as bad. What was happening now? Was it the fact that now I was outputting 8-bit uncompressed 1080p 23.98 via a Kona 3 to an HDCAM deck? Was the output via HD SDI thru the Kona sending out a clearer signal with all the hot spots and ultra dark areas that are somehow chopped off when you output via firewire to the AJ-HD1200 deck? I couldn't explain it.
So we cancelled the output and tried to figure out what to do.
The engineer...my buddy...had a solution. To export the footage in a special way. That way will be a secret that I will keep to myself. When I re-imported that footage and dropped it in the timeline above the originals, I found that the exported clip had chopped off the portions that are above 100IRE, and clamped (chooped, whatever) the portions that go below black. In other words, it does what Broadcast Safe should do, but doesn't really do.
So....I will do this process and try the output again tomorrow. The audio sounds GREAT though. Can't wait to watch it on TV. Sure, I have seen it 3 times already, and will watch it 2 more times...but I will still watch it when it airs. Or, rather, I'll TIVO it and watch it when I can.
5 comments:
Shane, I think this is happening because AFAIK an export Using QuickTime conversion goes through RGB which would explain why any subblack or superwhite levels are getting clipped off.
That is nice to know. Usually I just do something because I know it works...but never really know why. I guess that I don't NEED to know why, but it is nice to know.
It sounds like you were running into problems due to oversaturated highlights and shadows. This can be spot checked by turning on the Saturation option in the Waveform monitor (right-click and choose Saturation from the shortcut menu). This is the same as setting a hardware WFM to FLAT, displaying chroma saturation overlaid onto the luma. This will show the saturation sticking up over 100 and down under 0 on the scale (even though the Vectorscope looks fine, this is a different measurement). The tricky fix is to then use the Desaturate Highs (or Lows, they both do the same thing) filter to reduce saturation in those areas, while leaving the saturation of your midtones alone.
The biggest issue I had was not using external scopes that would see all of this. I just couldn't afford them. But now with FCP 5.1.2 out and the scopes looking better, I will be able to correct properly.
I think the most important piece of equipment is the hardware waveform and vectorscope. Don't use broadcast safe filters or software waveforms. Grade, grade, grade. Nothing should go to broadcast without a proper grade...that must be built into the budget. I enjoyed reading your blog.
Post a Comment