Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
Any comparisons between stock and hacked GH2?
  • Hey folks.

    I just wondered if anyone had put up any stills, or clips of the same scene between a stock and hacked GH2 where the improvements are clear to see?

    I've yet to hack mine and am about to go away for a week's holiday with it next week, I'm torn as to whether to hack with stable settings or just leave it as is because i'll be away from my computer so unable to revert anything if i encounter an issue.

    My gut says play it safe and wait, but I just wondered if there were any comparisons where the difference is so large that it'd be worth me hacking to try and improve any footage i get.

  • 49 Replies sorted by
  • I don't have anything to compare yet. i have lots of "before" stuff from an indie movie shoot but nothing from "after" yet. Honestly, from my testing, the quality increase in bitrate only is noticable but is *not* a linear increase with the increased bitrate and the subsequently larger video files. The biggest difference is in the dark areas of the frame where MUD would show up. On higher bitrate files, you see less mud in those areas but not a lot more detail in bright areas.

    In other words, diminishing returns.

    I think the magic bullet is going to be the quantization stuff VK and Cbrandin are looking into and not super high bitrates.

    I'd leave your GH2 stock. Who wants to worry about corrupt files and such while on holiday?

    You'll be plenty happy I think.

  • @svart

    Just remember to note that this is true if you talk about 24H.
    But difference is noticeable in motion or in high detailed scenes.

    As for stability, look for stable settings topic. We already have much more stable function comparing to GH1.
  • @jimtreats

    Here are the samples showing default firmware/32 mb/42 mb/42 mb 3&6 GOP. There are differences in the 1080p files but not as significant as some people have expressed. High motion 720p can really benefit from a higher bit rate since it is so low for the default firmware. However, B-frames and the quantization work should really optimize it.

    The 32 mb/sec settings in the thread below have no adverse affects. The camera will operate just as if it was new but at a higher bit rate.
  • @Vitaliy

    Are there any examples that show MUD or compression artifacts even with moderately higher bit rates? I haven't seen any evidence that higher bit rates really are necessary even with moderate motion.
  • I just did a quick test with the Lpowel seta.
    Made 4 shots of an silk dome tweeter, witch has very fine detail.
    To me the shot 1080 24 @ 42mbit looks the best, better then standard GH2.
    mjpeg 1080 looks as sharp as the 720P @ 42mbit
    The standard 1080 24 looks better then the hacked 720P and hacked 1080mjpeg

    With Lpowel seta only the bitrate is changed in avchd, tomorrow I want to test the setc from Kae.
  • The problem is that everyone just says “This looks better to me”. That isn’t telling us anything . You have to compare the exact same image for the settings side by side. Let’s see some screen shots of the differences. It is really easy to do a screen shot. Just use VLC and let it save it as a png file.
  • @jimtreats I can't tell you all differences. But this is typical path for those who apply hack.
  • @mpgxsvcd

    I'm not a tester, but hasn't Vitaliy said the difference is most noticeable in shots with motion? So will a screen shot be instructive? Seems like we need full motion video to assess.
  • @brianluce

    You really need both stills and video. There are some scenes like trees blowing in the wind that can produce artifacts without very much motion at all because they contain so much detail. Stills are perfect for those.

    For scenes with high speed motion and faster shutter speeds stills can work also. You just have to keep the motion fixed and consistent which isn’t always easy. I used a metronome for the scenes I shot. The motion is always the same.

    I have found that the ISO settings determine the bit rate more than anything else. If you shoot at ISO 3200 indoors it will always shoot towards the highest limit of the bit rate regardless of whether there is high detail or high motion. If you shoot at ISO 160 it will almost always be at the lower end of the bit rate range.

    I just find it interesting how everyone says that the high bit rate low GOP settings produce the best results without doing any systematic testing. Let’s face it the stock GH2 has been producing unbelievable results all along. Just because the shots look great with the high bit rates and Low GOPS doesn’t necessarily mean that the settings are beneficial. Saying that the new extreme bit rates are better without a one to one comparison is naive.
  • @mpgxsvcd , By the submission that noise is easier to see with the higher bitrate -This discovery alone tells us that the finer details (such as noise) and other finer image details are now being rendered visible.

    I have two GH2 cameras, one with the hack, the other stock. This weekend I will set up some low light shot comparisons and display them here.

    BTW, I just read your above post. Yes, I do agree with you. As it's been mentioned here, the GH2 at 1080 24P Cinema mode does produce a good image on its own. I do believe it has improved using higher bitrate in low light detail and motion such as pan movement and high detailed moving images such as water.

    My real question is these two scenarios:

    1. low light video shot at low ISO with little detail such as flat walls. With a low ISO setting, the codec will not be forced upon a higher bitrate. This may still cause less detail in the shadows and possible mush.

    2. low light video shot at low ISO with lots of fine detail. The fine detail in the video will increase the bitrate; therefore, the low light footage should have better detail and less mush in the shadows.

    Ideally VK and Chris and all other gurus will figure out how to increase the lowest bitrate which will force a higher average bitrate under the same conditions.

    Great work to all here man!

  • I have an example I shot of pebbles in a stream at 720/60P, in which compression artifacts could be seen in the water. I've not been able to shoot the same scene (was away for the shoot) but I'm going to try to recreate it with the new, higher bitrate.

    I'll try to post the original file so you can see this.
  • @proaudio4

    Yes there is a difference. Between the stock and modified bit rates. No doubt about that. Especially in noise rendering. You can see it in the pictures I posted. The stock firmware will simply smear the noise out. This may or may not be desirable depending on what you are looking for.

    The real question is how is this smearing affected by higher bit rates. There has to be a point at which the bit rate is ample to render an accurate image of the noise as it was recorded by the sensor. The problem is that we cannot determine where that point is without doing very systematic testing.

    I have done that testing up to 42 mb/sec and I believe it shows that there really isn’t much gain by going over 32 mb/sec. It is really hard to tell which picture is which for the higher bit rates. I will also post the rest of the videos to Vimeo tonight(I ran out of upload cap so I couldn’t post them last week).

    There is no need to do the tests with two separate cameras. Testing a camera with nothing checked but the version compare is sufficient to test the default settings. In fact it is better to test with a single camera fixed in one spot on a tripod.

    When I ran my tests I never moved the camera at all. I simply swapped the SD card out every time I updated the firmware. I didn’t even remove the legacy 55mm Canon FD lens in-between tests. The GH2 accepted the new firmware even without an m4/3s lens on it.

    I really would just like to see someone else show the differences they are seeing instead of talking about how great the results are. Vitaliy has been asking for this all along but no one seems to listen anymore.
  • @greeches

    We will need to reshoot the scene at the same time. It doesn't do any good to shoot it on different days from different angles. The actual differences are so slight that you really need to isolate all of the other parameters to do a fair comparison.
  • @mpgxsvcd

    I agree. I need to find a local spot that I can find the same type of motion and potential artifacts to test.
  • What about your Kitchen Sink? Just film the water running with a slow shutter speed and then use a high shutter speed. The flow rate is extremely consistent and reproducible.

    I will post the video of a fan that I shot the other night. It had a rotating imbalance and produced a very hard to compress scene. However, I didn't have a computer with me at the time so I couldn't do any comparisons.
  • @proaudio4

    "1. low light video shot at low ISO with little detail such as flat walls. With a low ISO setting, the codec will not be forced upon a higher bitrate. This may still cause less detail in the shadows and possible mush."

    I did this scenario and it produced the exact same bit rate regardless of the settings used. If the camera doesn't determine that it needs a higher bit rate then it will act just like stock even with the hack applied. This is where Vitaliy and Chris's work will probably show the most benefit.
  • Gotta push the limits of GH2 to see noticeable difference.

    Deep DOF on complex objects. e.g. Maple tree. There should be no mud.

    Fast moving object with deep DOF. Less mud.

    Black area details in low lighting. Raise ISO. 1250 or higher. More natural looking noise.

    Not clipped highlight. Try overblown highlight scene. I don't know. Is there any difference?

    Not sure about shallow DOF.
  • @ KrisCanonizado

    You have to keep everything fixed. Varying focus, ISO, shutter speed can all skew the results. It appears that your focus was not accurate with the default settings. That would be a much more significant change than the bit rate would be.

    Honestly, I don’t think this test can be done accurately without a true manual focus lens. Even if you use an m4/3s lens in manual focus mode it will reset itself between shoots. You have to make sure nothing changes at all between shoots.
  • @mpgxsvcd Yes, I do agree. I also would like to see tests done on the two scenario/s.
    I mentioned above.

    I will also post some footage where I accidently shot 720 60P (6 GOP) at 35mb/s. I thought I had selected 42mb (6 GOP).

    This 720 60P 35mb/s footage shows someone on a skateboard in lower light with a fence in the background.
    The skateboarder image falls apart in motion. It's very strange. His arms look like they are bent.

  • @mpgxsvcd " did this scenario and it produced the exact same bit rate regardless of the settings used. If the camera doesn't determine that it needs a higher bit rate then it will act just like stock even with the hack applied. This is where Vitaliy and Chris's work will probably show the most benefit"

    Thanks mpgxsvcd. I had a feeling this may be the case. I hope Vitaliy and Chris can find a way to increase the lowest bitrate.
  • I did something similar to this using the "cat of death" which routinely broke GH1 codec settings. My cat has very strange fur patterns and his fur has multiple layers of light and dark on each hair which make for a pattern very close to the pattern of death chart.

    I shot two files of his back while he was sleeping and compared the light/dark areas both in and out of the overhead light.

    While I don't have the files anymore, the outcome was this:

    24H stock and 24H GOP3/65mb(Kae's setting) looked similar in both the lighted and non-lighted situation. The 65mb setting had slightly less muddy dark areas but 3x the filesize. That's about it.

    I did the same thing with a tree outside:

    24H stock showed slight macroblocking in the dark areas of bark and shaded leaves. 24H with 3GOP/65mb(Kae's setting again) there was less macroblocking in both the bark and the shaded leaves but there was no discernible enhancement in detail in the well lit portions.

    I'm also of the opinion that finding a way to increase the lower bitrate limit will improve the areas of problems that I've seen. VK has mentioned a number of times that the 24H codec is much improved from the GH1 24p. I think this means that we will not see the drastic increase in quality from bitrate increases that we saw on the GH1. We'll see minor increases in quality in raising the upper bitrate only.
  • I think anything of high detail and movement will stress the codec; therefore the high bitrate will always be better. Especially if you're shooting the seaside.

  • I tried it out myself, if you shoot low motion, stick with standard bitrate if youre shooting 24p, but if youre smart, you can configure your cam to shoot in standard 22mbits on 24L and higher one, like 42 or 65mbits at 24H, so you can decide on location.

    High motion, especially at high shutter speeds with no motion blur benefits highly from bitrate increase, also from low gop.
  • @fatpig "High motion, especially at high shutter speeds with no motion blur benefits highly from bitrate increase, also from low gop.

    Good point fatpig!