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.
GH4 - Best Video Settings
  • 753 Replies sorted by
  • If you meter for bit 235 being white, then you have headroom. I agree that you don't get any extra bits in total or any benefit in ability to capture the available dynamic range. I have done my own tests and in terms of overall image quality there is very little difference. One of the things people often overlook is that the 16-235 range was an adaption of older NTSC analog signal spec's that allowed for NTSC analog black setup and headroom for analog tube imaging devices. Quite a lot of broadcast network delivery spec's still use the 16-235 range, even with file based delivery.

  • 4k photo mode uses 0-255 because it is photo quality standard. Why not shoot video too with 256 shades vs 219 shades if the best quality is the goal.

  • Because it still records the full range. It just sets metering for black & white levels differently. Bit depth is a bit more complicated than 'shades'.

  • If one of the goals with this camera is proper exposure and ETTR, shouldn't we be setting the custom curves to -5 highlights and leave the shadow at 0 to avoid additional noise? This should allow either more lighting be used or a bump up in ISO to bring out shadows a little better.

  • If you shoot a proper grayscale test chart you will see EXACTLY what the highlight custom curve settings do without guesswork. They do not change the clipping threshold, just the curve below that point. After a lot of testing I shoot Natural profile all settings at zero. Now I spend more time working on precise exposure and I'm getting superior results. BTW I grade everything in Resolve and use waveform monitors to set levels. Until Panasonic can provide a proper log curve profile for the GH4 I'm staying away from the Cine profiles as they just don't grade as easily as Natural. A lot of the custom controls are really meant for creating images which are not going to be graded in post-production. In that context, using curves and other settings is fine. For grading they can sometimes cause problems which I would rather not have to deal with.

  • caverport, On what exacly settings you are recroding? I see that natural but what about rest? Do you use MP etc. I want to test it by myself.

  • @bart_w The settings you may be looking for are sharpness, saturation, etc. I leave them all at 0. Master pedestal I always leave at 0. I sometimes use the highlight curve at -2 in bright conditions. I always use H264 MOV format.

  • @caveport

    @shian I'm interested in why you say video levels other than bit depth settings of 0-255 are nonsense. 16-235 is standard for most professional video cameras and is the internal processing standard for Avid systems. IMO it is provided to enable direct compatabilty with certain software apps so that the bitdepth range is not rescaled on import, as this can sometimes create artefacts in the image. It also allows a bit of headroom for highlights as bit 235 is set to be 100% white. Of course it's important to remember that these ranges are for 8 bit signals.

    You answered your own question with that last sentence. Nowadays, "16-235" is merely shorthand for the old 8-bit DV standards rather than actually specifying the bit-depth of the software tools we use to process video footage. Unless you insist on running your video editor strictly in 8-bit mode, your H.264 files will be decoded from 4:2:0 YUV and converted into 16 or 32-bit RGB format. Any scaling done from that point on will be calculated in far more precision than the original recording. It will likewise be scaled in high precision to match any export format you choose, with no need to match import and export formats.

  • @LPowell Good point. I still use the 16-235 GH4 setting after a lot of testing as I prefer to have the camera metering treating 235 as 100% white. It's not about what's best, it's about operational practice, and having a little headroom on highlights.

  • @caveport I'm new to this discussion and you probably know a lot more than I do, but using the 0-255, 16-235, 16-250 is not all about just to map the quantization from the sensor/internal processing to those range limits? If so, using anything but 0-255 will limit the “dynamic” range of values. Bomb dropped, away I go… :)

  • Using 16-235 does not clip off the lower and upper values, it narrows the scale of where it places them.

    Neil

  • @jopereira, the limiting factor of the GH4 is never luminance range, but bit depth and bit rate for 4K encoding. 100mbps 4.2.0 8-bit is rather thin for any luminance range at 4K resolution.

    If I'm not mistaken, the GH4 always encodes it's sensor data to a 16-235 range and only adds a tag whether to expand to a different luminance range after that encoding.

    Cheers, Pete

  • @c3hammer After a lot of testing I found that the camera encodes to the range you choose in the camera settings. It's provided to avoid remapping of levels by whichever editing software you may be using. As I mentioned earlier, Avid uses 16-235 internally as that is the SMPTE spec for YUV signals. Avid will re-map the 0-255 levels on import. Other more modern software may or may not use this range internally. I use 16-235 because I do a lot of broadcast delivery work. For web delivery or other non-broadcast jobs on FCP or Adobe editors, 0-255 may be a better option.

  • @c3hammer If we study 100% sensor readout the sensor itself is a limiting factor for a perfect image. The de-dayering gives unperfect color resolution even in RAW photos at 100%. The sensor noise is also a limiting factor. The GH4 4:2:0 100Mbs is surprisingly good and comparable to camera JPG.

    http://www.dpreview.com/forums/post/54849802

    http://www.dpreview.com/forums/post/54849358

    4k photo mode uses 0-255. It gives normal full 8bit range JPG photos. It the same for video.

  • @caveport That's what makes sense: camera output is in the range we choose, so I don't see the disadvantages of recording at 0-255 (best possible range, 8-bit) even if software/editor works in a different range. As said before, 32bit software (like Premiere, AE, etc.) make conversions (virtually) lossless - 32bit is great for detail (that's 1/4.294.967.296 precision!!), so every time you apply a filter to an image it's going to be to that scale (0-255) and not to a 16-235 narrower scale.

    The output of edited image can be anything you like, but all the worflow was made in greater detail.

    I'm not a specialist, just 'guessing' from what I understand.

  • @jopereira In the end it makes very little difference to the final quality. However the editing software does not convert to 32 bit. That is the internal processing bit depth when calculations are made for effects and other processing. In the case of Adobe, you are essentially correct, but with Avid, importing and processing is quite different. As I stated previously, the GH4 options are there for a reason. If you don't need them, fine. In my years of working as an editor and colourist, guessing is a bad option. No offense intended, all smiles here!

  • I am using GH4 video 0-255 luminance levels.

    It is quite confusing because some players shows only normal 16-235 levels and some editors does not use 0-255 properly. Some say 16-235 is better for codec (less information to process- more bitrate to details) etc etc. Graphic cards has full range/limited range (16-235) settings...

    If I watch my GH4 0-255 videos with Potplayer I must set level settings: input 0-255, output 16-235 to see all details in highs and lows with GH4 0-255 file.

    Why must I use output 16-235 while my monitor is still 0-255 RGB monitor? The logical way should be input 0-255 and output 0-255.

    Is the video signal always finally 16-235? Is there any benefits of using 0-255 in camera?

    The odd thing is that the graphic card or tv seems to use always 16-235 video and then convert it to monitor. It seems odd that I must send limited range video to video pipeline even if I have already full range video in a file. Is it true or is there a way to watch full range video without first compressing it to limited range and then again recompressing it to monitor range (RGB 0-255). Does the graphics card recognize when it is showing video and changes color processing accordingly? Can I ever have my 255 levels of luma in monitor?

    If the luminance (Y) is 0-255, what range are colors. I think that colors are in YUV format two coordinates (U,V) of color space but are these numbers 0-255 or 16-235?

    How can I watch my videos so that all the color data is transferred to my monitor or tv without needless conversions. Is there a way to see "GH4 RAW 0-255" video in monitor? Is there some "mode" or trick to send "RAW" video data directly to monitor via 3d features or via other methods? Potplayer has many options to show and send video data.

    Quite confusing.

  • @Vesku The metadata in H.264 files include a Full-Range Flag that when set indicates 0-255 luminance levels. Fully-compliant H.264 decoders and players should recognize this flag and act accordingly. In 32-bit display mode, computer graphic adapters and monitors all work natively with 0-255 RGB ranges. However, some graphic adapters have video overlay hardware that may be internally configured to work in YUV Rec 709 16-235 range. This can confuse software players that expect the standard 0-255 RGB monitor range.

    In standard Rec 709 format, legal 8-bit YUV range is 16-235 Y, 16-240 UV. In a Full-Range H.264 file, 8-bit YUV range is 0-255 YUV. In order to watch a video on your RGB monitor, the H.264 YUV data must be converted into RGB format, which is always full-range. There is no such format as "GH4 RAW 0-255"; that would be an 8-bit decimation of the sensor's 12-bit RGBG color array, which is not available in video mode.

  • @LPowell

    Thanks for very good information.

    "GH4 RAW" is my funny word for GH4 0-255 mode but it has in theory much more information than normal 16-235 video. In practise things are more complicated. When converting there may be all kind of interpolation or dithering which mixes results. Then there is sensor and codec unperfect behaving.

    I just wonder if there is IQ loss when watching GH4 0-255 video. I must adjust player or video card to convert GH4 0-255 YUV to 16-235 to see it properly. I wonder if these conversions use some floating point systems like better video editors so that all the imformation is converted properly. I cant see visually much difference between GH4 0-255 or 16-235.

  • @Vesku

    I must adjust player or video card to convert GH4 0-255 YUV to 16-235 to see it properly.

    This is a software compatibility problem rather than an interpolation or data precision issue. For best results, YUV formats should always be converted directly into RGB for display on computer monitors. It is never advisable to scale YUV formats from 0-255 into 16-235 8-bit format or vice versa.

    Full-range H.264 0-255 YUV is not a mode exclusive to the GH4; it is a standardized ISO format. It is used to record H.264 video by Canon and Nikon cameras as well.

  • @LPowell

    It is never advisable to scale YUV formats from 0-255 into 16-235 8-bit format or vice versa.

    If I edit a final video for general use video I must always convert 0-255 to 16-235.

    But why my players or video card must do that before I can see the whole dynamic range of 0-255 video. Can you see 0-255 video without any additional settings or tricks and without clipping? How?

  • @Vesku

    If I edit a final video for general use video I must always convert 0-255 to 16-235.

    Details count, I said "scale" YUV 0-255 to 16-235, not "convert". Unless you are working with an obsolete 8-bit YUV video editor, you are editing your video in (hopefully) 16 or 32-bit RGB mode. When you export a final video file, you are converting from RGB mode into YUV 16-235 Rec 709. That should not involve direct scaling of YUV 0-255 data into YUV 16-235 format.

    Outside your video editor, your video players must interact with your OS and video card driver in whatever fashion they can manage. It sounds like there are configuration mismatches in your system that result in improper display of Full-Range H.264 files. If your software can't be configured to handle these files properly, it may be easier to just record in 16-235 H.264 mode. That's what that GH4 option is intended for, to provide software compatibility options to allow you to match the sometimes inflexible requirements of legacy software and hardware systems.

  • @LPowell

    ...you are converting from RGB mode into YUV 16-235 Rec 709...

    Thanks for clarification, I actually ment that.

    Every computer I have seen works just like mine with GH4 0-255 files (must set output 16-235 in player or in video card). Has anyone done it differently?

  • I solved the levels issue for me.

    I compared the original GH4 0-255 video source frame and the same frame captured from monitor after output level adjustment 16-235 in Potplayer and video card full range 0-255 mode. I studied the frames in Photoshop and those are absolutely identical, no IQ loss or level conversion.

    It is quite disturbing and against common sense to set video output to limited range 16-235 before sending it to video card. The Potplayer has a floating level space and when using a limited output range it actually sends the right levels to video card and monitor. Nothing is lost.

    So I use in Potplayer level settings input (source) 0-255 and output 16-235 and video card full scale 0-255. Then I see all the original GH4 0-255 levels right in my monitor.

    I can have the same effect by selecting a limited range (16-235) in video card and leaving levels untouched in player. The practical thing with Potplayer is that I can alter levels if I want to enhance levels while playing files or if watching different cameras which has different levels. GH2 for example records 16-255.

  • For the purpose of grading is it better to shoot 0-255? Would the bitrate quality suffer slightly since now it's using the full 8-bits of each octet?