GH2 Incoming Canon shooter: I think I want a GH2, but I have a question.
- Thread starter HansSteinert
- Start date
Sage
Well-known member
Because it included a gamma shift option, it has been used as an automated color correcter to bring GH footage into a cinematic gamma space. Though - the real claim to fame for high end use is an algorithm to *recreate a smooth 4:4:4 10 bit color space from the sub-sampled h.264 4:2:0 8 bit space, as well as a workflow component to mass convert dslr formats into common post codecs (prores, dnxhd).
HansSteinert
Well-known member
How would you do that manually?
There are plenty of other programs for transcoding to ProRes that do it much faster & I don't think any of the evidence from various reviewers supports the proposition that sprinkling fairy dust over 4:2:0 8 bit colour space video to magically turn it into 4:4:4 10 bit colour space video is useful.
Justyn
Veteran
@ KHOLI... Thanks for all that you do for us and this site!!!! I haven't investigated the Hack or the RGB converter but I will on my next narrative project. I've been cursed and blessed by being so busy that I've been doing it the old fashioned FCP way just cause it works and I've not had the time to investigate another workflow.. but that's coming on my next break. Admittedly, I think that through the hack and through this alternate workflow I'll get a more pleasing image, but straight out of the box the GH2 is killer.... So awesome. I just wonder why we have to justify it all the time with canon people? And that there's not more of a buy in.. Invariably I'll have to show my camera and footage to Canon guys and they will laugh and scoff, that is until their camera is overheating... And I have a 5D and had a 7D and have used the T2 T3 canons and EVERY single one of them overheats on me. I'm a director who shoots and shoots and I don't break till I get what I want. I learned so much from working with Frank Oz that I'm now a changed man... I don't break the momentum and I shoot in blocks of time and its so much more productive... BUT the canon cameras CAN not handle it. Every single time they overheat... 5D not as much, but it still does and the 7D is the worst... well maybe the T2 is that...
Jaime Valles
Veteran
Thanks Kholi and everyone else for all the info on this thread! I'll be testing all of this myself over the next couple of days and posting results. 
Matthew P
Well-known member
I don't know how this compares to what everyone else is doing, but I've had great success lately using Sony Vegas' colour curves in its 32bit editing mode. The highlights seem to pop out of nowhere!
I've come to conclude that the GH2 actually has a much better dynamic range than I previously thought. The in camera processing must "squash" the shadows and the highlights to add more contrast, whilst not actually clipping. Just like adding more contrast with curves, I guess. Anyway, here's an example...
One thing I've noticed is that, because of the compression, some artifacts become visible in the highlights. Easily solved by filming in Cine mode, though - looks absolutely fantastic! I'm much happier with my GH2 now after finding how to get the DR to seem visually greater.
I've come to conclude that the GH2 actually has a much better dynamic range than I previously thought. The in camera processing must "squash" the shadows and the highlights to add more contrast, whilst not actually clipping. Just like adding more contrast with curves, I guess. Anyway, here's an example...
One thing I've noticed is that, because of the compression, some artifacts become visible in the highlights. Easily solved by filming in Cine mode, though - looks absolutely fantastic! I'm much happier with my GH2 now after finding how to get the DR to seem visually greater.
Sage
Well-known member
Very easy - just find the 'Rgb curves' option available in your nle, and alter the luma line into an s curve. The colors can further be altered to reduce overly 'hot' video wb - like tungsten/cfl reddish scenes; this is a bit more involved than the luma & is best done with good sense & constraint. No blue/orange bleeders or it'll be a kids picture
http://desktopia.net/wp-content/upl...RS-3-Shia-LaBeouf-As-Sam-Witwicky-575x323.jpg
S-curve:
http://199.89.180.250/~stars/images/fig-1.gif
Last edited:
Aki_Hartikainen
Carbonite Member
I don't know how this compares to what everyone else is doing, but I've had great success lately using Sony Vegas' colour curves in its 32bit editing mode. The highlights seem to pop out of nowhere!
I've come to conclude that the GH2 actually has a much better dynamic range than I previously thought. The in camera processing must "squash" the shadows and the highlights to add more contrast, whilst not actually clipping. Just like adding more contrast with curves, I guess. Anyway, here's an example...
One thing I've noticed is that, because of the compression, some artifacts become visible in the highlights. Easily solved by filming in Cine mode, though - looks absolutely fantastic! I'm much happier with my GH2 now after finding how to get the DR to seem visually greater.
This is important, because according to Kodak handbook for sensitometry, highlights are more important to filmic look than shadows. Film can not be overexposed, more and more silver will continue to accumulate as amount of light increases. This means that film has eternally more latitude for highlights than video. Now, Lumix cameras seem to only output values between 0 - 100 on a similar scale, whereas a Canon HV30 I had could output values to 108. When this scale is logarithmic, Canon seems to have more highlight latitude. However, for video delivery those 108 Canon brightness values need to be compacted to the scale of 0-100, so a camera could do that also. And this seems to be the case with GH2, as you have demonstrated.
I have had similar results using Shadow/Highlight filter in Adobe programs to recover some highlight detail. Clearly the GH2 highlight latitude would need to be examined to create a standard method to capture the most highlight latitude. Like this rather filmic sky from Canon HV30: http://www.fourseasonshd.com/cgi-bin/csvsearch.pl?db=db2&tp=tp3&search=1089
Aki_Hartikainen
Carbonite Member
That is a bit different thing and is separate from the 0-100 scale thing. Values sometimes called "super-white" and "super-black" exist, and it is up to the camera manufacturer to decide whether to use super-white areas and save them in a file. Canon HV30 saves them in a file up to 108, whereas Lumix cameras save files in 0-100 compacted or we could also say normalized.
The end result saved for video delivery must be normalized within 0-100, so I had to use curves and ProcAmp in Adobe Premiere to pull down super-whites recorded by Canon HV30 to video legal. Otherwise clipping would have occurred when saving to DVD, for example.
Danielvilliers
Veteran
I did a test on the gh2 versus the canon 7d dynamic range below
http://www.personal-view.com/talks/discussion/1795/dynamic-range-test-between-gh2-and-7d/p1
My conclusion is that the gh2 is at least as good as the Canon cameras with Marvel cine or cinestyle. The reason is because in video the gh2 has two tech that negates any of the Canon advantages. The first is because of its down-scaling which will affect the noise and in the process dynamic range. While the Canon is line skipping to what most consider a 1 to 1 pixel, the gh2 is sampling perhaps not the whole sensor but much more pixel than the final 1080p size. The second tech is the hack. With the high bitrate more or less CBR codec, the details in the shadows are being recorded while most of them in the Canon are discarded in form of macroblock and mush. As you can see in my link you can raise the shadows quite heavily in case of the hacked gh2.
The last thing, unfortunately on the net you always read the same people that repeat the same thing, like bigger sensor gets better DR etc. Using the same level of tech from the same generation yes. But that is not the case when you use different manufactures and generation of technology. For example the latest Sony exmor sensors have been better and better in DR compared to the Canon ones. The result is that the 1/2 size apsc Nikon D7000 trouce the full frame Canon 5D mark2 by a good 2 stop. For sure the latest full frame exmor Sony sensor in the D800 will be even better than the D7000 one. Lastly just look at dxomark and sensorgen reading between the Canon and GH2 and you will see that there is not much more than half a stop DR difference to start with.
http://www.personal-view.com/talks/discussion/1795/dynamic-range-test-between-gh2-and-7d/p1
My conclusion is that the gh2 is at least as good as the Canon cameras with Marvel cine or cinestyle. The reason is because in video the gh2 has two tech that negates any of the Canon advantages. The first is because of its down-scaling which will affect the noise and in the process dynamic range. While the Canon is line skipping to what most consider a 1 to 1 pixel, the gh2 is sampling perhaps not the whole sensor but much more pixel than the final 1080p size. The second tech is the hack. With the high bitrate more or less CBR codec, the details in the shadows are being recorded while most of them in the Canon are discarded in form of macroblock and mush. As you can see in my link you can raise the shadows quite heavily in case of the hacked gh2.
The last thing, unfortunately on the net you always read the same people that repeat the same thing, like bigger sensor gets better DR etc. Using the same level of tech from the same generation yes. But that is not the case when you use different manufactures and generation of technology. For example the latest Sony exmor sensors have been better and better in DR compared to the Canon ones. The result is that the 1/2 size apsc Nikon D7000 trouce the full frame Canon 5D mark2 by a good 2 stop. For sure the latest full frame exmor Sony sensor in the D800 will be even better than the D7000 one. Lastly just look at dxomark and sensorgen reading between the Canon and GH2 and you will see that there is not much more than half a stop DR difference to start with.
Aki_Hartikainen
Carbonite Member
Thanks for the test, I will have a better look in a moment. Just to make sure that we are talking about the same thing, I'll illustrate what I meant earlier. For Canon HV30 material I was able to expose the image more and later recover some of those highlights. The scale ends at value 100, but pulling down the levels revealed detail from the overexposed values. This gave me more highlight latitude. If I try to do the same with Lumix cameras, nothing above value 100 exists, when pulling down the levels. Image below.
If I just dropped the Canon HV30 footage on the timeline, I might not have known the whole sky was not entirely overexposed. I had to use levels to see to whether there was information not visible without adjustment in the high lights. This is different from Lumix cameras. They hard clip at value 100. I don't need to check whether there is information above value 100. Everything has to be within 0-100 already in-camera.
If I just dropped the Canon HV30 footage on the timeline, I might not have known the whole sky was not entirely overexposed. I had to use levels to see to whether there was information not visible without adjustment in the high lights. This is different from Lumix cameras. They hard clip at value 100. I don't need to check whether there is information above value 100. Everything has to be within 0-100 already in-camera.
Last edited:
righteous.
Well-known member
FINALLY somebody else gets it. Every time I try to discuss this people view me like the village madman. Specs aren't consistent between manufacturers!
Aki_Hartikainen
Carbonite Member
Well, I need to admit that I can not seem to be able to get such romantic skies as easily now with Lumix cameras as I did with Canon HV30. Had the lens not been so very poor in HV30, I might have started to investigate how to activate the HDMI output of HV30. But the lens was rubbish in HV30, so I switched to cameras where I could use any lens. Truth is, I would be happier with a combination of HV30 and GH2, the high light latitude would come from HV30 and sharpness and interchangable lenses from GH2.
Allow me to link to a sample of a romantic sky from HV30 with no color grading at all: http://www.fourseasonshd.com/cgi-bin/csvsearch.pl?db=db2&tp=tp3&search=1061
I just pointed the camera there and it captured the sky like that. I have lost appetite doing landscapes with Lumix cameras, because they do not have the same skies.
Allow me to link to a sample of a romantic sky from HV30 with no color grading at all: http://www.fourseasonshd.com/cgi-bin/csvsearch.pl?db=db2&tp=tp3&search=1061
I just pointed the camera there and it captured the sky like that. I have lost appetite doing landscapes with Lumix cameras, because they do not have the same skies.
Last edited:
Aki_Hartikainen
Carbonite Member
I had a look at your test and I noticed that you had used a spot meter to determine dynamic range of the cameras based on the reflected scene illumination. This is inaccurate. The camera does not necessarily see changes in brightness at 1/6400 in that scene. The spot meter only tells that value to be the correct exposure for that illumination, but that could still be vastly over exposed for the camera at 1/50, for example. There is no information in the lamp highlight, it is at maximum exposure and quite certainly beyond.
If you wanted to measure the dynamic range of a camera, you need to start lowering over exposure and determine the point where the camera starts to see changes in brightness. Same for the shadows, you start from pitch black and start to either open aperture or raise light wattage and determine the point where the camera starts recording changes in brightness. The difference between those points is the dynamic range of the camera for the used settings.
You have measured the dynamic range for the scene, not for the camera.
If you wanted to measure the dynamic range of a camera, you need to start lowering over exposure and determine the point where the camera starts to see changes in brightness. Same for the shadows, you start from pitch black and start to either open aperture or raise light wattage and determine the point where the camera starts recording changes in brightness. The difference between those points is the dynamic range of the camera for the used settings.
You have measured the dynamic range for the scene, not for the camera.
Aki_Hartikainen
Carbonite Member
Here I have compared two different film modes of GH1 (the shadow part):
We can see that the two film modes are nearly identical when it comes to capturing shadow detail. Differences in the curves start to show only at aperture f3.5. This would be accurate method to compare shadow detail of two different cameras also. Knowing the exposure where the camera starts to record shadow detail is the lower end of the dynamic range.
We can see that the two film modes are nearly identical when it comes to capturing shadow detail. Differences in the curves start to show only at aperture f3.5. This would be accurate method to compare shadow detail of two different cameras also. Knowing the exposure where the camera starts to record shadow detail is the lower end of the dynamic range.
DPStewart
Veteran
I don't know if this applies to what you're working with, but I have always achieved MUCH better skies on every single camera I have ever owned (Nikon, Canon, Olympus, Fuji, Lumix) by putting a simple polarizing filter on the lens. I actually thought this was sorta mandatory for outdoor stuff.
maranfilms
Veteran
Not to change the subject but this comment is total nonsesne.
Granted it may not be as sharp as some of the glass your using on a dslr, but by no means was the lens rubbish. In fact it was actually quite good for a 700.00 camera. My xha1 exhibits more C.A than my hv30. There's not many wide angle lenses that don't have barrel distortion. It's the nature of the beast. I kind of grew fond of the diamond shape flares I could get out of the hv30 lens. None the less a great cam it was and still is.
Granted it may not be as sharp as some of the glass your using on a dslr, but by no means was the lens rubbish. In fact it was actually quite good for a 700.00 camera. My xha1 exhibits more C.A than my hv30. There's not many wide angle lenses that don't have barrel distortion. It's the nature of the beast. I kind of grew fond of the diamond shape flares I could get out of the hv30 lens. None the less a great cam it was and still is.
Aki_Hartikainen
Carbonite Member
I understand, but I did not use a polarizer when I captured this with a Canon HV30: http://www.fourseasonshd.com/cgi-bin/csvsearch.pl?db=db2&tp=tp3&search=1089
I have browsed through the old material I filmed with a HV30, especially landscapes with strong skies and seas, and it is clear that HV30 has the better high light latitude. The skies are more detailed and vivid, as are reflections in the water. Less banding too. Canon made that right. This could be the reason why some people think Canon is the more "filmic" camera, but I only have experience in Canon HDV camcorders. Their lenses were poor. Chromatic aberration was unbearable. That is the reason I switched cameras. I gained something, but lost something too.
I have browsed through the old material I filmed with a HV30, especially landscapes with strong skies and seas, and it is clear that HV30 has the better high light latitude. The skies are more detailed and vivid, as are reflections in the water. Less banding too. Canon made that right. This could be the reason why some people think Canon is the more "filmic" camera, but I only have experience in Canon HDV camcorders. Their lenses were poor. Chromatic aberration was unbearable. That is the reason I switched cameras. I gained something, but lost something too.