Any GH2 news on the HDMI output front?
- Thread starter Caesar
- Start date
Hi Henry,
When you shoot greenscreen, do you shoot greenscreen with the camera body rotated 90degrees to get better resolution?
In your test, Could you show keys with the body horizontal, then same subject with camera body rotated 90 degrees, and then both scenarios with ETC for us pixel peepers....
Thanks for your efforts!
Hi Henry,
I downloaded your MTS file ( still downloading the giant MXF file ), and noticed two severe glitches in the video at 00:00:11:23 and 00:00:27:23. Can other people check to see if they are seeing the same glitches as me. ( I'm using Sony Vegas Pro 9 / 10 )
I downloaded your MTS file ( still downloading the giant MXF file ), and noticed two severe glitches in the video at 00:00:11:23 and 00:00:27:23. Can other people check to see if they are seeing the same glitches as me. ( I'm using Sony Vegas Pro 9 / 10 )
Phil Rhodes
Well-known member
Yes it has interlacing coming out of it. Yes it is possible to write recorder software which would automatically decimate it. I and others will be more than happy to do exactly that once someone comes up with the R&D funding.
Don't hold your breath.
P
Don't hold your breath.
P
I'd love to know if the HDMI issue is something that "in theory" might be fixed by a hack? Or better yet, in light of Phil Rhodes' comment above, something a company like Atomos could write into the Ninja?
I just sent Atomos an email asking about GH2 support.
I just sent Atomos an email asking about GH2 support.
Last edited:
farfromsubtle
Well-known member
It doesn't need a hack. As Sandbox was able to determine from my footage, the pulldown can be taken out through a combination of manual and automatic removal. It's laborious though, and he still can't get a grasp on the pattern (correct me if I am wrong Sandbox).
The point is, all of the footage is there- no skipped or interlaced frames once it has had surgery. I think it will be a matter of time before the ClipWrap, Cineform, Voltaic, ect, ect people come up with an automated solution.
The point is, all of the footage is there- no skipped or interlaced frames once it has had surgery. I think it will be a matter of time before the ClipWrap, Cineform, Voltaic, ect, ect people come up with an automated solution.
farfromsubtle
Well-known member
I don't see why not. All of the footage is there. Unless there is some sort of randomness factor to it that makes it impossible to do on the fly, then the more likely solution will be a post-processing workflow. Thing is, the Ninja could still capture the interlaced frames and spit them out for whatever app comes up with the solution.
I mean this in the nicest and most polite way:
I can't believe there was ever a question in your mind (other than for size/form factor). If many of us could afford a red and the accompanying equipment there would be no question we'd get the red. I can't even come anywhere near affording a 5Dmkii whose picture I prefer enormously to the GH2 and I currently work for a smaller hollywood studio and am about to transition to a top flight post/finishing/trailer vendor. I am beginning to wish I could just get a 60D but I think that extra 300.00 is too valuable right now to give up.
Enough of my complaining though, go get a RED!!! Godspeed!
Jon
Jim Arthurs
Well-known member
I see the same issues with the MTS file, viewed both in Vegas 9 and VLAN player on the desktop.
I've tried twice on two different computers to pull down the MXF, but both times it stops downloading and leaves a partial file around 500megs.
Henry, if you have time to re-record a clip using the same settings that's a bit smaller (say 200 megs) I will personally hand deliver it to the Convergent Design offices and go over it with the appropriate engineers.
I would suggest using a much lower quality setting on the data rate, as we're not trying to judge quality, just trying to predict a cadence pattern. Go as low a rate as I-frame will support... 100Mbit is fine. Or long Gop at 50Mbit/sec.
Do as someone suggested in an earlier post and turn off the menu feature that has the camera search for a Panasonic TV over HDMI.
Also, don't actually record to the on-board SD card while recording to the nano, just to see if the process of recording on-board over-taxes something and messes up the signal.
I personally think the odds are slim that this will be doable, but if you're willing to help, I'm glad to do what I can.
Regards,
Just received my PAL GH2 and have been doing some 50i testing using the Intensity Pro.
So far 50i is rock solid, no weird frame skips or pauses. (only tested 2 minute clips so far, will do longer tests later)
24p is a bit trickier and has some strange frame ordering to fit it in 50i (which is obviously a non-standard pulldown that won't be detected by any editors).
It's driving me mad going through each frame manually, so I am trying to write a utility to detect the frame order in the hope that some kind of pattern emerges. (at least something better than the 60i madness!)
We'll see what happens.
Martin
So far 50i is rock solid, no weird frame skips or pauses. (only tested 2 minute clips so far, will do longer tests later)
24p is a bit trickier and has some strange frame ordering to fit it in 50i (which is obviously a non-standard pulldown that won't be detected by any editors).
It's driving me mad going through each frame manually, so I am trying to write a utility to detect the frame order in the hope that some kind of pattern emerges. (at least something better than the 60i madness!)
We'll see what happens.
Martin
Are you able to restore progressive frames from the interlacing? aemonk has a PAL GH2 and he was feeding it to an AJA card. However, I could never extract any truly progressive frames from his footage. If you have a moment, could you upload a 10 second clip from your Intensity capture of 24P? The more investigating this stuff, the better.
I'll be surprised if you find a pattern. I've already gone through 2+ minutes of NTSC GH2 Intensity captured footage (thanks to farfromsubtle), and the only pattern I could find was that there was no pattern. Sometimes, the cadence from AA to BB (for example) would happen with just one stutter. Other times, the footage would advance from AA to BB (for example) and then fall back to AA, carry forward a short number of frames, and then repeat that stutter, then settle in. Other times, it would repeat that stutter another 4 or 5 times.
Then, there's the double frame that happens every 800-1000 frames or so. But even that is not consistent.
Check out this portion of this thread: http://www.dvxuser.com/V6/showthread.php?231315-Any-GH2-news-on-the-HDMI-output-front/page15
I would hazard a guess that the final stream cadence has more to do with buffers and cadence than some pre-programmed cadence designed to just keep everything smooth. This mess looks very random on timing, random on stuttering characteristics, and even random on frame doubling.
I still think the best strategy is going to end up being editing in 29.97, and then running the final product through a script that sets each clip up for a quick visual conform. By clicking a couple times through each clip, cadence can be determined very quickly, and move on. 3 seconds a clip, each clip averaging 3 sec, an hour show could be conformed in an hour or so. I could actually settle for that, considering the price of entry. Just gotta build the generic script.
I'll be surprised if you find a pattern. I've already gone through 2+ minutes of NTSC GH2 Intensity captured footage (thanks to farfromsubtle), and the only pattern I could find was that there was no pattern. Sometimes, the cadence from AA to BB (for example) would happen with just one stutter. Other times, the footage would advance from AA to BB (for example) and then fall back to AA, carry forward a short number of frames, and then repeat that stutter, then settle in. Other times, it would repeat that stutter another 4 or 5 times.
Then, there's the double frame that happens every 800-1000 frames or so. But even that is not consistent.
Check out this portion of this thread: http://www.dvxuser.com/V6/showthread.php?231315-Any-GH2-news-on-the-HDMI-output-front/page15
I would hazard a guess that the final stream cadence has more to do with buffers and cadence than some pre-programmed cadence designed to just keep everything smooth. This mess looks very random on timing, random on stuttering characteristics, and even random on frame doubling.
I still think the best strategy is going to end up being editing in 29.97, and then running the final product through a script that sets each clip up for a quick visual conform. By clicking a couple times through each clip, cadence can be determined very quickly, and move on. 3 seconds a clip, each clip averaging 3 sec, an hour show could be conformed in an hour or so. I could actually settle for that, considering the price of entry. Just gotta build the generic script.
Yes, there is actual progressive frames in the interlacing. I.E AB BC etc.
I tested this by taking two sequential interlaced frames and extracting the even fields of the first and joining them with the odd fields of the second frame, the end result was a perfect progressive image.
I pointed the camera at a black and white scrolling bar on a computer screen at 50p as the test footage.
The next step is to find a pattern, I will let you know the results.
I tested this by taking two sequential interlaced frames and extracting the even fields of the first and joining them with the odd fields of the second frame, the end result was a perfect progressive image.
I pointed the camera at a black and white scrolling bar on a computer screen at 50p as the test footage.
The next step is to find a pattern, I will let you know the results.
Just a thought, as someone has brought up in another thread.
As AVC bit rate isn't an issue on HDMI output, can someone test recording FSH at a shutter lower than 50 for PAL and 60 for NTSC. When recorded to AVC this yields slightly split frames i.e. AB BC CD DE etc...., but the pattern is fixed and can be deinterlaced into progressive frames.
Is it worth testing?
As AVC bit rate isn't an issue on HDMI output, can someone test recording FSH at a shutter lower than 50 for PAL and 60 for NTSC. When recorded to AVC this yields slightly split frames i.e. AB BC CD DE etc...., but the pattern is fixed and can be deinterlaced into progressive frames.
Is it worth testing?
auzgc - that's wonderful. I just checked aemonk's footage again - and I still can't create progressive frames from it no matter the order and time-offset I use to combine fields. Perhaps he wasn't shooting 24 fps on camera? It's been a while. But good to see progressive over HDMI isn't limited to NTSC.
Also interesting that you're having success using . . . wait for it . . . the Blackmagic Intensity!!!! It's the only digitizing system we've seen preserve the 24P cadence through HDMI. (Of course, our sampling is really small, but it will be good when someone succeeds using a non-intensity card to do any of this). So far we've seen attempts using AJA, Nano, and Matrox - to no avail. It could be camera issues or user issues, but there's a trend so I merely point it out.
Also interesting that you're having success using . . . wait for it . . . the Blackmagic Intensity!!!! It's the only digitizing system we've seen preserve the 24P cadence through HDMI. (Of course, our sampling is really small, but it will be good when someone succeeds using a non-intensity card to do any of this). So far we've seen attempts using AJA, Nano, and Matrox - to no avail. It could be camera issues or user issues, but there's a trend so I merely point it out.
HKO
Well-known member
I was realistic in my expectations for sure and knew the whisperings that it may have been crippled. Just so you know - The GH2 was never my 2011 serious camera purchase - I am saving for that at the momennt and don't really know what it will be - probably a RED Epic S. But I have done some projects with HDMI capture using an adaptor and a consumer camera and the clients were thrilled - so HDMI capture for £ 800 with the advantage of no cumbersome adaptor, interchangebale lenses, good low light etc would have been really good.
By the way guys - if you don't press record on the camera and just record the interlaced stream it truly is a stunning image anyway. Progressive is a pain but interlaced does rock - and it has its uses. There is the glitch every few seconds of course.
I will record a few more tests tonight with a friend of mine that looks way prettier than me.
Later
aemonk
Active member
I really tried every single shooting mode. Since I'm now a little more familiar with the GH2, I'll give it another HDMI out try to the AJA io HD. I'm also now on CS5. Maybe Premiere Pro interacts different with that AJA device than FCP does ...