Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
GH4 4K Panasonic video camera, User reviews and opinions
  • 1273 Replies sorted by
  • I borrowed that same cam to do more tests, and I think he got a lemon... everything I shoot in 1080 looks like crap.... noisy, blocky...it's awful. Could it be just the MOV and MP4 formats? I haven't tried the AVCHD but it's my understanding that the AVCHD is an even lower bitrate.

  • No 8x8 matrixes. Yes 2x2 4x4 matrixes we need hack!! We need 10bit to SD card we have 5.1 profile h264 codec is capable of more... Yes maybe processor cant...but could try!!!

  • Nevermind - He had upgraded to ver 1.1 and didn't reset everything - I did a global reset on the camera and it looks great! But not liking how Reds get blocky at 2 stops under, and muddy as fuck anywhere below that.

    Still playing with the pedestal and curves settings. and bouncing around with 16-235, 16-255, and 0-255. So far I'm finding these settings take a little getting used to... they make it really easy to underexpose the camera if you're not using a meter, and then not paying attention to the in cam meter...

    When properly exposed this camera is very nice. If my testing shows me that there is 11-12 claimed stops (which likely means 9 actual and then 7 usable) then I'll order mine tomorrow. So far the shadow response is as disappointing as the BMCC. But the highlights behave very nicely.

    When I have concrete data and video to go with it... I'll post it.

  • @shian - wow so is it usual practice to reset globally after installing new firmware? Just upgraded mine but it will take considerable time to customize camera again after reset. Do you think this issue was particular perhaps just to this camera? Have others had this problem after latest firmware update? I guess better to be safe than ruin a shoot!

  • only if you have issues... If i'd known he'd upgraded the firmware and I was having these problems it would have been the first thing I did

  • "So far the shadow response is as disappointing as the BMCC" . Shian

    Good day, Herr Shian. Can you post a link or example that shows what you mean on the BMCC. Do you mean the image falls apart when you raise the shadows? Or is there just no info in the shadow?

    I know you'll be doing gh4 test when you get to it.

  • @chauncy 2 things - image gets pretty blocky on certain colors once you get below 2 stops under even on 200 ISO (just like the GH2 [even hacked]) and can't be lifted without artifacting, I would have thought with a baseline 100mbps design (a la DVCPRO cameras) they'd have a better shadow response like the HVX had.

    On the BMCC you can't really lift the shadows much - but you can dial in the highlights... I'm finding the same here with the GH4. I will continue to test, and will post results likely in a few weeks.

    A lot to like about this camera....a lot. After shooting with it for a few days and then doing some side-by-sides with the GH2 Intravenus I was not liking the way my old GH2 operated and felt in my hands and on my rig... I NEVER thought I'd say that.

    I'll postpone a final opinion until I see what I can really do with it in terms of settings, falloff, DR, and then in post - different looks... some with crushed shadows, others with lifted shadows, and various contrast ratios. Wish I had more time to immerse myself with this cam. I think it might have really great potential without an external recorder, but too much going on right now to do that.

  • thanks, Shian

  • can anybody post a hacked gh2 and a gh4 comparison?

  • I'll try to work on getting something shot fairly soon with luck. It's funny though, every time I look at the footage from either one after looking at the other I'm always impressed. Haha... They are both great!

    The 4k footage is simply amazing though, and I think it's a bit cleaner than the GH2 footage. 4K sounds unnecessary until you squeeze it down onto a 1080 or 720 sequence, and then you see it is a lot more detailed than the 1080 footage.

  • I don't have time to export and upload the video, but here's some stills from an extreme high contrast/low light test. The 3rd one is showing the amount of sensor crop there is when shooting 4K...quite a lot. A Speedbooster is likely next on my list of things to buy, cuz wow it crops the hell out of an already cropped lens circle.

    Light Setup - 100 watt CFL light against the wall, and a diffused 100 watt CFL dome light spill from hallway to the left about 17 feet away. Handheld meter reads ambient light in the room at an average of f1.4 (7EV) @1600 ISO. The hotspot on the wall reads 7 stops over. The side of Spawn's cowl where we get detail back is 4.3 stops over. Where the detail on the pale yellow wall starts to pick up detail is 3.3 stops over. The corner of the towel that is edging out of the light and into the shelf shadow is even exposure and the top left corner of frame is 4 stops under.

    If you study the light splash, especially to the left of the image, you'll see where the detail and color starts to return more gradually with the GH4. Also notice the color retention on the towel and the guitar headstock.

    In the GH2 shot, at around the 9th fret or so, the neck on the guitar has a highlight on it , cuz the door was cracked slightly further open so part of the spill from the hall was falling directly on it and bringing up the exposure on that left side of the frame a bit.

    If you discount the hot spot which is blown out, this shot has a DR of around +4 and -4 (9 stops) but 3.3 on top before pale yellow blows out and 4 on the bottom giving us about 9 stops with detail. I'd hesitate to call it 7 usable stop just yet until I do a skin tone test. But I think I'll find 7 usable and 9 with at least a little detail giving us a true 11 stops of DR on the cam. (1 being black, 11 being pure white) Which is what film had up until about 15 years ago when they found ways to expand it. That should be plenty.

    GH2 Intravenus 1600 ISO, 20mm Pancake, f1.7 [smooth -2,-2,-2,-2] image

    GH4 4k downconverted to 1080 1600 ISO, 20mm Pancake, f1.7 [CineD, curves preset 2, pedestal 0, 16-255] image

    crop factor - this is GH2 HD cropped to meet framing of 4k GH4 image

    Much better falloff in the GH4 image in the shadows, the grain even at 1600 is nice and smooth, oh and the detail... you can actually read the titles in my DVD collection with the 4K. Also notice the amount of shadow detail in the light wrap around the shelf and the lamp - considerable improvement there. I can shoot a lot more of the things I like to shoot with this cam, without having to make it work in post... I can get it in cam a lot easier.

    What I did in Post on these: I had to desaturate the GH2 footage to match, cuz it was much more saturated which at this ISO created a bit of color smearing compared to the GH4. Since the GH4 was shot at 16-255, I dropped the black levels down to broadcast to match the GH2, otherwise you'd see a noticeable shadow lift which gives one a false sense of shadow response. I've found that 0-255, 16-235, 16-255 are all kind of nonsense. They don't help your exposure or image quality at all... they only do for you what 5DtoRGB does to MTS files, move the black and white points around, they don't give you any more latitude. To test this; I shot the same thing with all three settings, then moved the appropriate white and black points to match the full 0-255 spectrum... they look identical. Then moved all the white/black points to match 16-235... same thing.

    GH2 with no desat image

    gh2.jpg
    1502 x 846 - 321K
    gh4.jpg
    1498 x 844 - 1M
    crop.jpg
    1920 x 1080 - 336K
    Screen Shot 2014-07-12 at 12.26.38 PM.png
    1500 x 846 - 2M
  • Just to answer my own question from a while back - as it seems like there are no yagh lovers out there. A 3s Lipo works great as a power supply for the yagh. We now have a handheld 4k shooting rig using a gh4, red rock shoulder rig (cheaper rigs are just as good), with shotgun and radio mic audio, HD monitor and led lighting options all on hot shoe mounts I was dubious about the yagh but it really is the business. And a 3s Lipo gives you all the mobile power you need at a knockdown price. Latest version has tidier wiring :)

    gh4rig.jpg
    1396 x 956 - 139K
  • @shian, great stuff! I've been meaning to do some more empirical testing to confirm some of what I thought to be true, but now I feel even less urgency! Ha... :) So are you going to shoot at 0-255 now, or did I misunderstand your conclusion?

    @glopro, what monitor are you using there? I have a cheap Lilliput 668GL that worked very satisfactorily with my Moon T7 GH2, but it does not work with the GH4... I have to decide if I want to get something else economical for the GH4, or wait until a combo monitor/4K recorder comes out for it and spend some bucks on that. Every once in a while I feel like the in-cam data rate is just a little bit shy of where it really needs to be for a shot here and there, so I'm mildly tempted by a clean HDMI recorder.

  • @Tjabo. It's a Lilliput 668 and we too are waiting for an external 4k recorder with monitor. We couldn't get the Lilliput to work at 24Hz only at 50hz system frequency.

  • @tjabo - haven't decided yet. Setting blacks to 16 does give you the ability to gauge how much blocking and noise you will have in your blacks when playing back the footage onto a monitor. So I may shoot 16-255. But it will require more testing for me to find a comfort zone that produces the best possible image. And to see if any of these settings stress out the camera and codec, and are to be avoided.

  • @shian Just so I understand your post correctly.. setting to 0-255 doesn't provide any additional luminance to work with in post?

  • Test with GH4 and Mitakon EOS to M43 Lens Turbo focal reducer.

    Test with GH4 and Mitakon Lens Turbo Eos to M43 focal reducer and Contax Zeiss 85mm F1.4 and Voigtlander 58mm F.4 lenses.

    Settings:

    Natural: 0,0,0, +1, 0

    Cinema 4k clips transcoded to ProRes 422(HQ) in Sorenson Squeeze Pro 9 then imported into FCP 7 on a PRORES 422(HQ) 1080P timeline sequence setting.

    i.Resolution-Off

    i.Dynamic-Off

    Master Pedestal-0

    Highlight/Shadow-0/0

    16-235

    Contax Zeiss 85mm f1.4 shot at F2 to F2.8(first 4 clips)-Contax to EOS adapter.

    Voigtlander 58mm f1.4 shot at F4.(last 3 clips)-Nikon to EOS adapter.

    POST DONE:

    RGB Balancing

  • On the subject of external monitors/viewfinders. Does anybody know of one that works?

  • @Soupertrooper

    I suggest to check out http://www.personal-view.com/talks/categories/monitors-viewfinders

    Topics for popular monitors usually have posts from GH and BM cameras owners.

  • @superset yup, as far as I can tell. But don't take my word for it. SHoot a shot with all 3, and test it out.

  • @superset, I did some tests with all the different luminance settings and profiles. I'm not exactly sure what the GH4 does differently, but I find the highlight roll off to be smoother and to show less clipping with 16-235 than 0-255.

    here's a gif animation that goes back and forth between the two showing the scopes from Vegas. I believe the scale for Vegas scopes are -20 to +120 IRE. It also doesn't do any adjustments based on the flags in the file. Vegas is a wysiwyg editor, unlike Premiere and FCP that automatically try to display what ever you shot in to 16-235. http://www.c3di.com/c3films/cineV-0_255-vs-16_235.gif

    Anyway, I'm finding my workflow to be easier with 16-235.

    Cheers, Pete

  • @c3hammer

    16-235 doesn't roll off the highlights, it just displays the luminance info differently straight out of the cam. It doesn't add any DR or treat highlights any differently in the recorded image. We all went through this with 5DtoRGB. These settings in the camera are all about what starting point you would like to work with without having to manipulate anything in post.

  • @shian, I have. If you look at the image, you can clearly see that 0-255 clips harder and is much more abrupt at the top of the highlights. 16-235 is a smoother roll off. I know I'm not the best colorist by any means, but I can't get anywhere near as clean an image with that hard clip in 0-255 when adjusting in post.

    I will admit the two scenes are not locked off at exactly the same time, so there may be some differences in the clouds and brightness. Definitely more of the clouds in the 0-255 clip so it's not definitive, but I've seen the effect a number of times with bright daylight scenes.

    Curious to see if others are seeing the same thing in their scopes.

    Cheers, Pete

  • Attached are shots of a color chart against white overexposed by 2 and half stops. One set to 0-255 the other 16-235 both on 4k, Cine D, curve 2, Ped 0. (the color depth here is 32 bit)

    I see what you're talking about.... there's some data extending up above 100 in 16-235.... but it's still hard clipping white, and there's zero difference in the upper left corner where the light is falling off.

    What it does give you is a +.01 gamma shift in 16-235. I added more pics to show that dropping the gamma in the 16-235 image to 99 makes both look completely identical whether set to 235 or 255. regardless of how they were shot

    I'll have to do a gradient test to see how the gamma shift affects highlight rolloff. But I doubt it'll make any difference.

    0-255.jpg
    2285 x 914 - 441K
    16-235.jpg
    2296 x 969 - 2M
    16-235 [255] gamma correction.jpg
    2295 x 986 - 491K
    16-235 gamma correction.jpg
    2295 x 986 - 491K
    16-235 [235]gamma correction.jpg
    2292 x 971 - 485K
    0-255 moved to 235.jpg
    2297 x 968 - 457K
  • @shian So, what's your take on the GH4 so far ... quick summation of specific usefulness/problems vis a vis other cams around ... and with what basic post?