Any GH2 news on the HDMI output front?
- Thread starter Caesar
- Start date
deadfish - should be the same. Use 60i. We're looking for green-screen footage and more chip charts - rotated at 45 degrees at some point would be nice. At this point, I think we all know what kind of mess we're dealing with (and to be fair, it's the mess we're choosing to be involved in - Panny never made any promises). It's more curiosity about subsampling and green-screen capability, etc.
For easy green-screen test, use your computer monitor to display a shade of green that matches the lighting on your desk - then a toy with hair or long fur can be used for very fast set-up. Using Photoshop to adjust the green brightness/darkness works as a separate lighting system. My fovorite subject for this is a "my little pony." Hah. Multicolored mane and all. Yes, I have 3 daughters.
For easy green-screen test, use your computer monitor to display a shade of green that matches the lighting on your desk - then a toy with hair or long fur can be used for very fast set-up. Using Photoshop to adjust the green brightness/darkness works as a separate lighting system. My fovorite subject for this is a "my little pony." Hah. Multicolored mane and all. Yes, I have 3 daughters.
Thank you Barry!
HKO
Well-known member
I reckon Panasonic have to be adding this extra nonsense in the code. I mean the AVCHD has the correct frame sequence and it is not difficult to send out the signal to the HDMI right? I mean what is the difference between 60 fps and 59.94 fps. Well if my maths is right that is a difference of 1 frame in every thousand right if the signal were progressive.
For crying out loud other cameras that have progressive sensors can send out a 59.94 fps signal. How hard is it to fix that - I mean every 17 seconds there should be a blip but not every few seconds. Okay the GH2 outputs interlaced fields but if you are extracting 24 progressive frames at the sensor level then the discrepancy jumps to 41 seconds. I don't know if my maths is right but my point is that the excuse is rather lame as to why this is happening.
So I imagine then that there is added code to actually scramble the HDMI signal........................ and added code can always be removed right? See where I am going with this..........? Vitaliy where art thou.:happy:
Currently uploading a green screen clip to my server - will take a few hours as it is over a gigabyte but will add the link here. It is a Nanoflash file direct from the CF card. The white balance was auto so the image has a yellow tinge from the lights - easily fixed if you have the tools. I use Cineform firstlight.
For crying out loud other cameras that have progressive sensors can send out a 59.94 fps signal. How hard is it to fix that - I mean every 17 seconds there should be a blip but not every few seconds. Okay the GH2 outputs interlaced fields but if you are extracting 24 progressive frames at the sensor level then the discrepancy jumps to 41 seconds. I don't know if my maths is right but my point is that the excuse is rather lame as to why this is happening.
So I imagine then that there is added code to actually scramble the HDMI signal........................ and added code can always be removed right? See where I am going with this..........? Vitaliy where art thou.:happy:
Currently uploading a green screen clip to my server - will take a few hours as it is over a gigabyte but will add the link here. It is a Nanoflash file direct from the CF card. The white balance was auto so the image has a yellow tinge from the lights - easily fixed if you have the tools. I use Cineform firstlight.
Last edited:
HKO
Well-known member
www.henryolonga.net/Nanoflash_green.MXF
Right click and save as. So sorry that I couldn't find any better models.
There ya go guys - like I said - over a gig in size due to the high data rate - you better have fast broadband. Recorded to nanoflash at 280 mbps I-frame. No pulldown removal attempted. Cineform can convert this as well as the latest versions of Vegas being able to edit them directly. YMMV according to your NLE. Badly lit green screen and white balance is a little off but you can get the idea of what the GH2 is capable of when recorded to a 35 Megabytes a sec codec. Big difference compared to the AVCHD IMHO. Why oh why did Pana have to do this .............pulls hair out.
Best regards
Right click and save as. So sorry that I couldn't find any better models.
There ya go guys - like I said - over a gig in size due to the high data rate - you better have fast broadband. Recorded to nanoflash at 280 mbps I-frame. No pulldown removal attempted. Cineform can convert this as well as the latest versions of Vegas being able to edit them directly. YMMV according to your NLE. Badly lit green screen and white balance is a little off but you can get the idea of what the GH2 is capable of when recorded to a 35 Megabytes a sec codec. Big difference compared to the AVCHD IMHO. Why oh why did Pana have to do this .............pulls hair out.
Best regards
HKO
Well-known member
Do you have an AVCHD clip to coincide with the nanoflash footage? Downloading it now - and thanks.
Uploading now
The link will be www.henryolonga.net/Nanoflash_green.MTS
The file is longer because I forgot to stop it on the Camera - duh. Anyway way smaller.
Give it a few minutes
Nope, HDMI output is not as simple to engineer as it may appear. It requires the camera to multiplex the video and audio streams together into a complex set of precisely timed digitized signals. Even if it didn't have to deal with the ludicrous mismatch between 60 and 59.94 fps, the camera's processor would need extremely fast wits to flawlessly handle sensor scanning, AVCHD encoding, SD card writing, and HDMI output all simultaneously in real-time. In recording mode, the GH2's HDMI output is intended solely for visual monitoring, and engineering it to meet recording-quality specifications would have no doubt increased the expense, risk, and development time budgeted to the camera.
HKO
Well-known member
I see your point of view but come on bro. This is a company that knows how to do hdmi. Let me put it this way - Panasonic know how to do this right. ALL their consumer camcorders obey the hdmi spec. This is a non spec output as far as the integrity of what is being pushed out the pipe - I mean what is it? 60i in-and-out-of-sort-of-up-and-down-in-and-out-cadence? Of course you can monitor fine and playback is okay but the output is technically out of spec. It's their call of course but I don't think it is any harder for them to get it right in a GH2 when every camcorder they have ever made does it right. Unless of course they simply don't want it to work. This is the ONLY camera/camcorder I have come across that does this - not suggesting I have owned them all but the ones I have used by big brands all work within spec whcih means that if they output it - I can record it and the signal behaves. My HFS10 had an EDID mismatch with the intensity but I got around that with a piece of hardware that set a firm EDID. No chance with this thing...............should I give Panasonic the benfit of the doubt? Perhpas but I am just cynical that they do not wan to eat into their higher end
The HDMI quality is actually stellar if it weren't for the issues we have discussed. It beats my HFS10 in detail which has '900' lines of res.
If it didn't produce such a stunning image anyway I would go on but this camera as is, is going to do a lot of work for me this year regardless.
The HDMI quality is actually stellar if it weren't for the issues we have discussed. It beats my HFS10 in detail which has '900' lines of res.
If it didn't produce such a stunning image anyway I would go on but this camera as is, is going to do a lot of work for me this year regardless.
Fohdeesha
Well-known member
Of course their camcorders are fine, they are camcorders. These are digital photography cameras, that happen to shoot video. It's a brand new market, and I'm willing to bet when they originally engineered and scaled out what chip they would need and implemented it, the thought of making sure hdmi output was perfected was the last thing on their minds. I'm sure they assumed it would be used by customers to hook it up to their tv at home to watch what they recorded earlier that day. They never could have forseen what they would end up being sought after for, and there is absolutely no reason to purposely cripple it. They want to sell as many of these as they can, trust me. I'm waiting on either picking up an f3, or waiting for a dslr to come out with decent dynamic range, s/n ratio, and normal, working hdmi output so I can just use it as an imaging head and handle recording and audio with a ki pro separately. I thought I had found it with the gh2 but it seems not 
The only problem with that theory, Fohdeesha, is that there is a whole new emerging game-changing market upon us, + recorder industry - KiPro, NanoFlash, Atomos Ninja, & another I can't recall the name - all of us waiting desperately for well over a year for a DSLR with clean HDMI. All the recorder manufacturers say hang on, they're coming. The firstest with the mostest will win. You think Panasonic isn't aware of this?
RBG
RBG
Maybe but I'm sure Panasonic has always known about the engineering & capabilities of some of these recorders which have been available for well over a year, and their proposed specs before that. It wouldn't take rocket scientists to figure out the ramifications of something proposed to be an HDMI recorder.
Thanks for posting those files, Henry.
However,
I can't remove pulldown from your MXF file. Of course, Shake doesn't recognize MXF, so I had to transfer it to ProRes. Used the 444 to be sure to preserve everything possible (overkill, of course). It's interlaced most frames, progressive some, and has huge tears across the screen every once in a while. It keys OK, but the interlacing really shows up, even with little movement. I'd post a screen grab, but these grabs are so small on DVXUser that they're not worth it. Perhaps someone else should take a look at the files and see what they think. See if it tears on their computer as well. Maybe that's why P Bloom never did the tests on his GH2/Nano combo. Maybe he couldn't get it to work. Hah.
This looks very similar to the footage that AEMonk was posting earlier in this thread. And, can I assume Henry, that your GH2 is an NTSC version?
Very strange. One thing that showed up was that it appears to be recorded at 30 fps (MXF). Perhaps try 29.97 or the HDMI spec for 59.94. An even number seems where all the trouble is to start with.
PS. Feeling a little grouchy about all this, but have to admit the keying blows smoke rings around the AVCHD version. Of course. The AVCHD key looks smoother because it loses all hair detail - which is NOT the goal here It also shows the lava effect - when 2 edges approach each other, they suddenly merge or "glop" together, forming a movement that is reminiscent of lava lamps. But what do you expect from 24 mbits of anything. Ugh.
However,
I can't remove pulldown from your MXF file. Of course, Shake doesn't recognize MXF, so I had to transfer it to ProRes. Used the 444 to be sure to preserve everything possible (overkill, of course). It's interlaced most frames, progressive some, and has huge tears across the screen every once in a while. It keys OK, but the interlacing really shows up, even with little movement. I'd post a screen grab, but these grabs are so small on DVXUser that they're not worth it. Perhaps someone else should take a look at the files and see what they think. See if it tears on their computer as well. Maybe that's why P Bloom never did the tests on his GH2/Nano combo. Maybe he couldn't get it to work. Hah.
This looks very similar to the footage that AEMonk was posting earlier in this thread. And, can I assume Henry, that your GH2 is an NTSC version?
Very strange. One thing that showed up was that it appears to be recorded at 30 fps (MXF). Perhaps try 29.97 or the HDMI spec for 59.94. An even number seems where all the trouble is to start with.
PS. Feeling a little grouchy about all this, but have to admit the keying blows smoke rings around the AVCHD version. Of course. The AVCHD key looks smoother because it loses all hair detail - which is NOT the goal here
It also shows the lava effect - when 2 edges approach each other, they suddenly merge or "glop" together, forming a movement that is reminiscent of lava lamps. But what do you expect from 24 mbits of anything. Ugh.
HKO
Well-known member
Hey Sandbox - yes it is a Japanese NTSC version. The nanoflash identifies the signal automatically. I do not think there is user intervention. Only to tell it to record psf or remove pulldown. Apart from that - the recorder determines the format. I could stand to be corrected if I am wrong.
Now - you do raise an interesting point though. What if they are streaming this in a 30 p container instead of 59.94? They cannot stream 60p of course and they don't do 24p native down the HDMI so perhaps they have chosen to do it in 30p but with a different EDID. That might explain the difficulty in getting down from 60p at the sensor to 30p to then contain 3:2 pulldown. I can see why the math would get screwed up. Only thinking out loud - dunno if it makes any sense.
Well I am moving on even if this gets fixed - probably to RED - they don't cripple anything LOL. In fact I had pre-oredered an AF100 until I saw how much harder it is to colour correct 4:2:2 longop stuff compared to RAW. I decided that in my heart of hearts I am tired of jumping through hoops to get a stunning image out the box. I want to shoot and than pull the card out the cam straight to edit into a hi res timeline. Only Alexa and RED allow that at the moment and I cannot afford an Alexa. I mean the really hi res stuff by the way without using an off camera recorder.
The GH2 will be my talking heads for corporate stuff camera - from a convenience perspective - it cannot be beat.
Now - you do raise an interesting point though. What if they are streaming this in a 30 p container instead of 59.94? They cannot stream 60p of course and they don't do 24p native down the HDMI so perhaps they have chosen to do it in 30p but with a different EDID. That might explain the difficulty in getting down from 60p at the sensor to 30p to then contain 3:2 pulldown. I can see why the math would get screwed up. Only thinking out loud - dunno if it makes any sense.
Well I am moving on even if this gets fixed - probably to RED - they don't cripple anything LOL. In fact I had pre-oredered an AF100 until I saw how much harder it is to colour correct 4:2:2 longop stuff compared to RAW. I decided that in my heart of hearts I am tired of jumping through hoops to get a stunning image out the box. I want to shoot and than pull the card out the cam straight to edit into a hi res timeline. Only Alexa and RED allow that at the moment and I cannot afford an Alexa. I mean the really hi res stuff by the way without using an off camera recorder.
The GH2 will be my talking heads for corporate stuff camera - from a convenience perspective - it cannot be beat.
I don't know what's going on. I've seen good recordings from Blackmagic Intensity cards ONLY!!!! (farfromsubtle on this thread) and y.kimura's vimeo piece. I only just noticed that.
Everyone else, for whatever reason, has interlacing over interlacing. Does anyone feel like this "simple" concept is turning more and more into a rabbit hole?
Henry's MXF file listed it's frame-rate at an even 30 fps for some reason. I wonder if the GH2 is, in fact, using EDID to change frame-rates based on what it sees or doesn't see. Doesn't explain what secret sauce Blackmagic is using. It is magic, though, I'm sure.
Either there's an EDID issue, or a PAL/NTSC issue, or a recording-device issue. It's leaning to the latter, in my opinion. AEMonk on the AJA card, yourself (Henry) on the NanoFlash, and the guy on Cinedigital.tv (http://www.vimeo.com/17776923) using the Matrox all shared similar mashed-up interlacing. The 2 who are getting clean (if you can call 7 seconds "clean") pull-down are using Intensity cards.
Can't wait to see what Barry has to say about his experience with this. If the Nano can't handle it, . . . well hopefully it won't take Barry as long as it's taking PBloom.
But as it stands now, GH2 looks to me like it's just a nice little addition to the VDSLR market, and better left at that.
Too bad.
Everyone else, for whatever reason, has interlacing over interlacing. Does anyone feel like this "simple" concept is turning more and more into a rabbit hole?
Henry's MXF file listed it's frame-rate at an even 30 fps for some reason. I wonder if the GH2 is, in fact, using EDID to change frame-rates based on what it sees or doesn't see. Doesn't explain what secret sauce Blackmagic is using. It is magic, though, I'm sure.
Either there's an EDID issue, or a PAL/NTSC issue, or a recording-device issue. It's leaning to the latter, in my opinion. AEMonk on the AJA card, yourself (Henry) on the NanoFlash, and the guy on Cinedigital.tv (http://www.vimeo.com/17776923) using the Matrox all shared similar mashed-up interlacing. The 2 who are getting clean (if you can call 7 seconds "clean") pull-down are using Intensity cards.
Can't wait to see what Barry has to say about his experience with this. If the Nano can't handle it, . . . well hopefully it won't take Barry as long as it's taking PBloom.
But as it stands now, GH2 looks to me like it's just a nice little addition to the VDSLR market, and better left at that.
Too bad.
HKO
Well-known member
I have an intensity and magma chassis here as well. Will record again later and see if I can get seven clean seconds as the other guys are. Will do this later. To be fair - seven second takes is not insurmountable if you are doing green screen stuff in studio. Portability becomes a real pain in the butt but for studio stuff.it may work.
Have a good one.
Henry
Have a good one.
Henry
farfromsubtle
Well-known member
Sorry I haven't been able to do the chromakey tests. Production is hectic right now and this is not in the front seat on priorities.