Yes of course. The concern is more on the unhacked codec falling apart on complex scenes
My take at matching the AF100 and GH2
- Thread starter Barry_Green
- Start date
M. Gilden
Canon DSLR Moderator
Exactly. There was a test floating around the web somewhere where someone did a hacked vs unhacked side-by-side, trying to prove that for simple scene with subtle movement the hacks don't necessarily provide any discernible difference.
A lot of folks seem to think the hack magically increases color depth and DR or something, when really it just preserves details that might otherwise be lost in compression.
M. Gilden
Canon DSLR Moderator
That's true. It is more gradable. But I think that has more to do with compression artifacts eating detail.
Maukka
Member
This thread was originally about matching on-camera recorded AF100 AVCHD footage with GH2 (hacked) footage.
And the results presented in this thread are very impressive. By far the best match anyone has ever seen before.
I use AF100 as my main camera and would like to use my GH2 as a B cam more than I have done so far. But I have not yet hacked my GH2, and after all, both cameras record internally in the same AVCHD codec which can be handy too.
This is why I am wondering if the settings presented in this thread should also give good results when using a non-hacked GH2 together with an AF100.
And the results presented in this thread are very impressive. By far the best match anyone has ever seen before.
I use AF100 as my main camera and would like to use my GH2 as a B cam more than I have done so far. But I have not yet hacked my GH2, and after all, both cameras record internally in the same AVCHD codec which can be handy too.
This is why I am wondering if the settings presented in this thread should also give good results when using a non-hacked GH2 together with an AF100.
M. Gilden
Canon DSLR Moderator
Short answer: Yes
Although the hacked footage will be easier to grade and tweak any discrepancies in post.
Barry_Green
Moderator
They will of course apply to a non-hacked GH2. But there is a fallacy in your statement of "both cameras record internally in the same AVCHD codec". That's not true, the AVCHD in the AF100 is more advanced and more efficient than the stock codec in the GH2. The GH2 can't get the same level of quality out of its 24mbps, as the AF100 gets out of its 24mbps. The codecs don't have exactly the same toolsets available to them. The hacking lets the GH2 have much more bandwidth, which lets it overcome its other limitations and allows it to exceed the 24-mbps-limited AF100's codec.
So they're not exactly the same, but even so, the purpose of my matching wasn't about the codec at all, it was about the tonal rendering, about the dynamic range, about the color (which still isn't quite to my level of satisfaction, but is pretty decent)... and those things will all apply equally to a hacked or unhacked GH2.
So they're not exactly the same, but even so, the purpose of my matching wasn't about the codec at all, it was about the tonal rendering, about the dynamic range, about the color (which still isn't quite to my level of satisfaction, but is pretty decent)... and those things will all apply equally to a hacked or unhacked GH2.
Maukka
Member
Thanks Barry!
I did think that the settings probably would indeed work just as well in non-hacked GH2. But as I know very little about the hacks and what exactly they do, I could not be sure.
It is brilliant that you guys spend all that time in researching and trying out settings and then share them here. Much obliged.
Thanks also for correcting what I said about the AVCHD codec on the two cameras. I was quite unaware that there are such differences between them.
Barry_Green
Moderator
Yeah, that was something that was actually disheartening in the GH1 -- it came out saying it was "AVCHD", but the codec was HORRIBLE. Ugh. Worst codec ever released on any video camera, ever. Worse than HDV. That's what made Vitaliy's original hack so brilliant, is it turned that awful so-called AVCHD into a real quality image. But I think it really put a stain on the name AVCHD, to have such a horrid codec using that name.
Although, here's the funny part, we used that camera (unhacked, as there was no hack at the time) to make a film called "A Little Mouth To Feed", which went on to be the opening-night red-carpet gala film at Slamdance 2009! So even though it had a hideous codec, you could still do good work with it.
The AVCHD in the GH2 (unhacked) is about 600,000 times better than what was in the GH1, by the way.
Although, here's the funny part, we used that camera (unhacked, as there was no hack at the time) to make a film called "A Little Mouth To Feed", which went on to be the opening-night red-carpet gala film at Slamdance 2009! So even though it had a hideous codec, you could still do good work with it.
The AVCHD in the GH2 (unhacked) is about 600,000 times better than what was in the GH1, by the way.
I'd have to differ with you on that assessment. The unhacked GH2 has a feature that in my view, makes it unreliable for professional use. In shooting situations that overload its limited bitrate, the unhacked AVCHD encoder may drop into a coarse, low-quality Fallback Mode that allows it to continue recording without interruption. The disturbing thing is that the camera can drop into Fallback Mode without any warning or indication to the user. That means that you have no assurance that your footage was recorded at the intended level of quality.
In the course of developing Flow Motion v2, I saw the unhacked encoder plunge into Fallback Mode on numerous occasions, even when shooting seemingly ordinary scenery. I documented one example here:
http://www.dvxuser.com/V6/showthread.php?283724-GH2-100Mbps-Flow-Motion-v2-Patch-Banding-amp-Blocking-Be-Gone!&p=1986152358&viewfull=1#post1986152358
One of my priorities with Flow Motion v2 was to insure failsafe operation by preventing Fallback Mode from occurring under any shooting circumstances. In my view, that is a crucial requirement for making the GH2 a reliable tool for production purposes.
Barry_Green
Moderator
Well, yes, I agree entirely, although I don't think you understand what I was saying exactly -- I'm saying the "fallback" mode still looks much better than the unhacked GH1! So it was (IMO) never worse, and almost always much better. (at least, that's the way I'm remembering it, but you have to understand I was very very displeased with the GH1's codec, so maybe I'm being harsher on it than it deserved? Not sure how that's possible though! 
)
Your work in the GH2 is vastly appreciated, and I use FlowMotion entirely. I know you know where you're coming from and what you're talking about. I'm just pointing out that my statement was intended more as disparagement of how just truly awful the GH1 unhacked codec was.
Also, frankly, I very much appreciate your dedication to reliability. Reliability (or the lack thereof) is exactly why I got rid of my hacked GH1s, because I simply could not count on them. And why it took me forever to bother hacking the GH2. And when I did, initially, I still wouldn't use it. Until I found FlowMotion and read more about your design goals with it -- that's what got me to finally hack it and keep it that way.
)Your work in the GH2 is vastly appreciated, and I use FlowMotion entirely. I know you know where you're coming from and what you're talking about. I'm just pointing out that my statement was intended more as disparagement of how just truly awful the GH1 unhacked codec was.
Also, frankly, I very much appreciate your dedication to reliability. Reliability (or the lack thereof) is exactly why I got rid of my hacked GH1s, because I simply could not count on them. And why it took me forever to bother hacking the GH2. And when I did, initially, I still wouldn't use it. Until I found FlowMotion and read more about your design goals with it -- that's what got me to finally hack it and keep it that way.
Thank you for your vote of confidence, I went through a similar evolution in perspective myself. Although I was able to get impressive results out of the GH1, it was never something I could completely rely on. After I started working on patches for the GH2, I soon realized that the reliability tests I used on the GH1 were much too lenient. The GH2's AVCHD encoder is not only far more robust than the GH1's, it's more hackable as well, and that allowed me to explore its limits more thoroughly. Fallback Mode is a hidden feature that's intended to protect the encoder from bitrate overloads. It was quite challenging to figure out a way to safely disable it!
OlegKalyan
Well-known member
Barry, lpowell,
thank you for your great work,
question, if you don't mind:
in this settings, how does the exposure value is similar from both cameras?
would the same exposure of 1/100 of a second at equal f stop on both cameras provide similarly exposed picture?
what would be the pedestal setting to match two cameras, at similar exposure pairs?
Thank you!
Hello from Moscow, we remember you Barry from your last visit here, which was very helpful!
thank you for your great work,
question, if you don't mind:
in this settings, how does the exposure value is similar from both cameras?
would the same exposure of 1/100 of a second at equal f stop on both cameras provide similarly exposed picture?
what would be the pedestal setting to match two cameras, at similar exposure pairs?
Thank you!
Hello from Moscow, we remember you Barry from your last visit here, which was very helpful!
Barry_Green
Moderator
Hi Oleg,
I used 400 ISO on both cameras, and set exposure via the waveform monitor and 100 IRE zebras. I don't recall looking at the lens for the exact f-stop, but I'm pretty sure it was basically the same for both.
I used 400 ISO on both cameras, and set exposure via the waveform monitor and 100 IRE zebras. I don't recall looking at the lens for the exact f-stop, but I'm pretty sure it was basically the same for both.
OlegKalyan
Well-known member
Thank you Barry, what do you mean saying that reducing pedestal to -9 diminishes certain abilities of the AF100?
Barry_Green
Moderator
It doesn't diminish any abilities, it just makes the image contrastier and crushes the shadow detail a bit more than I think is necessary. The higher the master pedestal, the more shadow detail you retain.
Your AF101 profile and demonstration are really interesting.
To me, i've already put this message in another very similar discussion, and before reading the first comments of this one, my best AF101 profile to match with GH2 (smouth,-2,-2,-2,-2) was
Gamma BPRESS
Knee Low
Matrix Norm1
Detail -3
V details -3
Coring -7
Chroma -3
Ph Chroma -3
Master Ped 0
DRS OFF
Make the manual white balance on AF101 and GH2 by pushing the butom (DO NOT use the same manual setting with x KELVIN because of very bad results). On AF101 don't forget to make the Black balance. And the images will match good.
On post, you 'll just have to reduce the hight luma (about 5 IRE) for the AF101 footage, with the level fonction, and may be the black level too...
I always set the Coring to -7, for all of my AF101 profiles, because of the best resolution by far. That means NO CORING. And details on -3 don't produce any problem. And then, It's very sharp !!
The GH2 produce the best resolution. But with this profile, the AF101 is very sharp too, and not far from the GH2.
Now, Barry, i've to try your configuration... I already see a RVB gamma difference (specially on faces), and about color, some red/magenta correction to do on the GH2 footage... isn't it ?
Well... try and see...
Gilbert (from Lyon-France)
To me, i've already put this message in another very similar discussion, and before reading the first comments of this one, my best AF101 profile to match with GH2 (smouth,-2,-2,-2,-2) was
Gamma BPRESS
Knee Low
Matrix Norm1
Detail -3
V details -3
Coring -7
Chroma -3
Ph Chroma -3
Master Ped 0
DRS OFF
Make the manual white balance on AF101 and GH2 by pushing the butom (DO NOT use the same manual setting with x KELVIN because of very bad results). On AF101 don't forget to make the Black balance. And the images will match good.
On post, you 'll just have to reduce the hight luma (about 5 IRE) for the AF101 footage, with the level fonction, and may be the black level too...
I always set the Coring to -7, for all of my AF101 profiles, because of the best resolution by far. That means NO CORING. And details on -3 don't produce any problem. And then, It's very sharp !!
The GH2 produce the best resolution. But with this profile, the AF101 is very sharp too, and not far from the GH2.
Now, Barry, i've to try your configuration... I already see a RVB gamma difference (specially on faces), and about color, some red/magenta correction to do on the GH2 footage... isn't it ?
Well... try and see...
Gilbert (from Lyon-France)
Last edited:
I did tests with your settings and I could not get them to match. I used manual white balance but the problem was exposure. But even with manual WB the shadows with the GH2 are still green while the AF100 is not.
The problem I found was that using cinema with the GH2 made much darker images than your AF100 settings. So I had to bump the ISO in the GH2. At ISO 200 in the AF100 I had to bump the GH2 to ISO 320 to look similar. I thought the exposure was matched? Or did I do something wrong?
But if I used Smooth with the GH2 the exposure matched your AF100 settings better. I actually have the impression that your AF100 settings match better the GH2 with smooth than with cinema. If you compensate the ISO then cinema is not too bad, but without compensating the ISO cinema didn't match at all for me. But smooth looked similar without ISO compensation.
Any ideas what I may have done wrong?
In my opinion -9 crushes the blacks way more than any mode in the GH2. In my test when the GH2 was still seeing details in the shadows with both cinema and smooth, the AF100 with -9 pedestal saw only black. It also makes blacks look nasty somehow. Like the shadows are dirty. I think -5 looks closer to my eyes than -9. Maybe -6, but -9 is too much. But if you found that -9 could be a better matches for the blacks, any ideas what I could have done wrong? I watched in 2 different monitors to make sue it wasn't the monitor.
I used the same lens with both cameras for the tests. The 14mm pancake lens at f2.5.