Ben Scott
Veteran
Right. But it's more bandwidth than 150mbs long gop and with the improved motion cadence etc that comes from intra frame recording. So I'll wait until then and see if log is worthwhile on this camera or just a marketing gimmick.
GH4 GH5 Owner / User Reports
- Thread starter ROne
- Start date
Right. But it's more bandwidth than 150mbs long gop and with the improved motion cadence etc that comes from intra frame recording. So I'll wait until then and see if log is worthwhile on this camera or just a marketing gimmick.
Peter Berry
Well-known member
This is just the nature of the DSLR/mirrorless full sensor binning/lineskipping down to a 2MP 1080 frame: the max res is in the neighborhood of 700 L/PH, but the GH4 ETC mode, which is a 1:1 small central crop to 1080x1920 for an overall crop factor of about 5X, gets to 800. See pg. 2 of Adam Wilt's very technical 3 year old GH4 review, where he documents a full 1080 L/PH res. on downscaled 4K - for a huge 50% res. increase over full-sensor 1080. 4K 1:1 native reached about 1600.
http://www.dvinfo.net/article/acquisition/micro-43/review-panasonic-dmc-gh4-camera.html
Last edited:
But will it improve the look of V-Log footage by taking IPB compression out of the picture ?
Bern Caughey
Veteran
Joe,
Where do you find this kind of information?
Does anyone know at what high frame rate (VFR) the image begins to degrade? Is it above 60fps, or higher?
Thanks!
Thomas Smet
Veteran
Different types of encoding and yes 400mbps is more but is it enough?
We don't have to look far to see how well the GH series has handled all-I formats. Both the GH3 and GH4 had all-I HD recording. The GH4 can record HD with 50mbps IPB or 200 mbps all-I. Care to take a wild guess which one looks better. The answer is typically the 50mbps IPB format looks better on the GH4. 200mbps was designed for 1080p 60p. That means 24p only gets to use about 80mbps of that. That is about what ProRes422 LT uses and while that is a decent format it can have macro blocking and artifacts.
There are two forms of compression going on with IPB H264. Image compression (like a jpg image) and temporal compression. The temporal compression helps bump up the image quality by tossing out blocks that are the same between frames. Works exceptionally well except when motion gets too complex and those predicted blocks can no longer be predicted very well.
All-I doesn't have those random motion blocks but it has some pretty heavy image compression b(again think a jpg image compression). It can have heavier image compression depending on the data rate because it doesn't save a lot of data through motion prediction of IPB. All-I can look amazing but it really needs enough bits to make that happen. What 400mbps vs 800mbps turns into is like a jpg photo compressed at 50% vs 75%. You don't have to worry about motion related artifacts but every single frame has image macro blocking, mosquito noise and other image based artifacts. The lower the data rate the more you will see these artifacts. Video also tends to show these artifacts more because the macro blocks and mosquito noise move or dance around from frame to frame.
The theory behind IPB is that you use the I frame at a higher quality so each frame can look much better than it should. It saves a lot of bits by not compressing each frame. When you do get macro blocking from complex motion the belief is that because of motion blur and how quickly those portions of the frame move, most people will not notice it as much. It becomes well sacrificed detail. IPB can look better in situations where there is little motion. Sit down interviews, most shots in movies with a locked down camera or slight dolly move, seminars and so forth. Heavy movement works better with an all-I format, subjects such as sports, surfing, explosions dance performances with a lot of fast moving people and lights.
So like Joe said it really depends on a lot of factors and at no time can anybody look at a bitrate number and say for sure a format will look better. We have plenty of previous experience with the GH3 and GH4 to realize all-I bit starved h264 formats never really do well compared to their IPB cousins. Some motion artifacts may be improved but what you are left with in the overall image isn't always a good tradeoff.
The 400mbps on the GH5 is designed for 30p. 24p will only use 320mbps of that because there are less frames. ProRes422 HQ is double that for 24p and even that is considered a highly compressed image format.
I'm sure the 400mbps will be a nice format but it will have artifacts and there will be better choices with the GH5 in terms of recording format. If one is absolutely concerned about quality then external 800mbps 4k is the way to go. Internal 400mbps will have different artifacts and may be very nice for a lot of people but don't expect a night and day difference over the 150mbps format. They will likely be comparable with a different set of tradeoffs.
Thomas Smet
Veteran
It does look like the HD is cleaner on the GH5 than it is on the GH4. Does not look like pixel binning from the samples I have seen online. I read/heard somewhere last week that VFR up to 120fps in cleaner than 180fps. I don't know if that is true or not but if it is I think it is more realistic to shoot 120fps and use optical flow in post at 2x to get to a virtually perfect looking 1080p 240fps. Optical flow on that fine of slow motion should turn out perfect since there is so little change between frames and you get the cleaner down conversion this way.
Thomas Smet
Veteran
Ok does anybody know if the GH5 loses the ability to manual white balance when in 10bit mode and using an external monitor/recorder? That is one of the most annoying things on the GH4. Any other functionality lost or does the GH5 work exactly the same when the HDMI is set to 8bit or 10bit?
I can work around these limitations on the GH4 but it is annoying.
I can work around these limitations on the GH4 but it is annoying.
Ben Scott
Veteran
As I have repeatedly said I will ignore log for now and will test if it is fit for purpose when the 400mbs firmware update is out. Because whether or not it is good enough down the line 150mbs long gop most certainly isn't worth shooting log in, in my opinion. I gave up on shooting log on the A7sII as well for not dissimilar reasons.
I couldn't really care less about what has come before in the GH5 range. This appears to be a nice little camera for rough and ready shoots where nothing more than a small mirrorless is required and is clearly different from its predecessors in many ways. I'll check out if it is any better quality down the line when updates have happened. I'm not really sure why this is such a controversial thing to say.
joe12south
Veteran
From Panasonic, but off the top of my head I can't remember exactly which source. You can also find some user tests on YouTube that confirm.
Not verified, but what I've perceived:
<= 60fps seems to be same sampling (of course hjgher frame rate changes compression requirements somehwat)
<= 120fps seems to drop quality
> 120fps seems to drop another level of quality
Even 180fps has more detail than 96fps on GH4.
joe12south
Veteran
We're simply saying that there is no reason to expect better PQ from the all-i option. A LOT of people have this expectation and don't understand why they shouldn't. If you find the h264 compression artifacts objectionable now, you most likely will then. To make a material difference, you're almost certainly going to need to use an external recorder with a different codec.*
*There's always a chance that Panny will surprise us and move to a better codec, but that doesn't seem to be what they have planned based on what they've communicated so far.
joe12south
Veteran
That's not the cause of the artifacts we all object to. The macroblocking color errors occur in perfectly still shots. It's more a function of scene type (flat colors and subtle gradations) than motion delta between frames.
joe12south
Veteran
No, it probably won't be more bandwidth per frame...for the average scene, all things being equal, it will amount to less. That's what we keep trying to explain. When you remove the efficiencies of long-gop encoding the bandwidth requirements to get similar quality skyrocket.
Ben Scott
Veteran
I have been specifically talking about the use of log and the nature of log as a large bucket it's only worth using if you are capable of filling up the bucket. Hence me saying I'll test it again after the update.
Thus far (one paid shoot in low key controlled environment ISO 250 Cinelike D sound recordist giving me a line in to the XLR adapter) I've zero complaints and quality wise it did everything I'd have wanted it to.
I will, however, go to my grave preferring intra frame recording codecs.
Ben Scott
Veteran
That is why, for the umpteenth time, I said I'll test it then and find out.
Thomas Smet
Veteran
Not sure. Even AVC-Intra and AVC-Ultra have issues with log material. What happens with these h264 encoders and log is the encoders are used to compressing normal video data and yet log has all those values crunched extremely close together. On detail they look great but in areas of flat subtle color changes you can get colors that are so close together the encoder assumes they are the same color. This leads to extra blocking or grouping of those colors instead of a nice subtle gradient. So flat areas like white walls, skies and large patches of human skin can show these artifacts with log. Normal rec709 formats don't have this issue because the encoder are getting values they expect to have. People also don't tend to push a rec709 nearly as hard as a log format.
I have tested AVC-Intra (HD 100mbps) vs ProRes422 HQ (HD 176mbps) and in areas of detail like grass the formats were extremely close but in areas of flat color the AVC-Intra had blocked together colors. This was with a log format that had to be graded to rec709. AVC-Intra is a pretty robust professional format and yet when it comes to log it can still have some of the same color blocking artifacts. IPB has nothing to do with this on its own and is really the nature of h264 or similar formats that just don't have enough bits for those flat areas.
ProRes422 HQ eliminates almost all of these potential issues by being a dummy format and throwing a lot of bits at each frame. Thats the only way to really handle log so the encoder doesn't try to get fancy grouping together what it thinks similar colors are. Every all-I format I have seen in that 400mbps 4k ratio have not been as good as ProRes422 HQ and looked more like ProRes422 LT. LT isn't a bad format but it isn't great either. Log brings out the worst in encoders so going with the best format you can will always help when shooting log.
With that said I do feel the 10bit 150mbps already greatly reduces the number of V-log artifacts on the GH5. I doubt the all-I format will make much of a difference. Having 10bit and an encoder that can better handle how to treat subtle colors has already done wonders for V-log to the point where even internal 8bit looks a lot better from what I have seen.
Thomas Smet
Veteran
Please do. All we are saying is don't expect miracles. It is easy to assume something that is 400 vs 150 will be better but there are dozens of factors that can cause that to not be true. Just don't get your expectations too high. Really who cares? The 400mbps is a free update coming for the GH5. Its a giant pot of gravy and if it does look better then awesome. If not no big deal because the 150mbps is already pretty darn good. Panasonic had enough stones to give us true 4k 10bit 4:2:2 HDMI output so we can choose what kind of format we need. If the 400mbps doesn't blow you away you still have plenty of other solutions to get what you need out of the camera.
SuperMachoAlpha
Member
I noticed that when recording 4k 60p there is a long delay when I stop recording. Is there any way around this or is it just me?
Same here
The GH5, finally, does not lose the ability to manually white balance, while in the 10bit mode. This fix was very much needed.
SuperMachoAlpha
Member
I was at a job this morning and I noticed it was taking way too long to stop and record. It's annoying.