GH5 Evaluating GH5 4K 10bit clips recorded at both 150Mbps and 400Mbps.

I am a wildlife photographer/videographer and occasionally I use 4k video cameras for purposes of extracting frames and converting them into photos. I have both the GH5 and Canon 1DX Mark II. Before vers. 2 of the GH5 firmware was available, I never would use the GH5 for fast motion filming of small birds-in-flight because IPB would occasionally spoil the shot. I would always use the 1DX Mark II for it's motion-JPG codec. I haven't tried the GH5 yet with all-I, but in theory it should work the same as MJPG.

This is a rather unique use of video, but occasionally it comes in handy. You can create some stunning images by overlapping frames to show the motion of flight.
 
Aajay said:
.... I never would use the GH5 for fast motion filming of small birds-in-flight because IPB would occasionally spoil the shot.
Click to expand...

I have not yet seen a spoiled frame in my GH5 4k 60P ipb videos and I have thousands of them.
 
Vesku said:
Here is GH5 1080P 100Mbs ipb vs 200Mbs all-i test (not mine). I cant see any difference in 400% enlarged examples. Panasonic ipb is very good.

https://www.youtube.com/watch?v=Xq8oTomWpeM&t=12s
Click to expand...

That is again not very complex shots that would typically break an encoder. Especially one encoding HD IPB at 100mbps which is a ton of bits for HD. You need a lot more complexity in the shots to see a difference. If this is typical of the type of stuff one shoots then without a doubt IPB can be just as good.

I think there is a significant disconnect when it comes to understanding what can break an IPB encoding at a given bitrate. The higher the IPB bitrate the more it will retain a solid looking frame with no macro blocking because there are more than enough bits to handle the entire frame. Thats why using 400mbps for IPB is kind of silly because after a certain threshold you end up with frames that are basically the same as 400mbps all-i. They are just calculated and built in an inefficient way. But visually you pretty much end up with the same thing. The entire benefit of IPB is to have a lower bitrate while yielding as close to the all-i quality as possible in most cases. It is those rare cases (exploding cars/buildings, tree branches/leaves blowing in a wind storm, tires kicking up a spray of dirt/mud, a bubbling brook/stream) when the "lower" bitrate IPB has to make sacrifices and duplicate blocks between frames. If the scene is not very complex or if there are a ton of bits to share between the frames then visually the IPB and all-i should look the same or close to the same. It is only when the bitrate is too low (10-25mbps for HD, mpeg2) that each frame no longer has enough bits to visually reproduce each frame without some crude duplicating of blocks of pixels between frames.
 
Thomas Smet said:
Thats why using 400mbps for IPB is kind of silly because after a certain threshold you end up with frames that are basically the same as 400mbps all-i.
Click to expand...
How so? I do not understand how that would be limited to your 400Mbps.
Why would a 400Mbps IPB not be roughly equivalent to 2 to 3 times the bitrate of an all-intra version, i.e. 800 - 1200Mbps?

IPB is a much more efficient compression technique because you can use information from neighboring frames while all-intra compression is restricted to information from the current frame only.
Sure IBP is difficult to edit but compression wise it is superior to all-intra.
 
Cary Knoop said:
IPB is a much more efficient compression technique because you can use information from neighboring frames while all-intra compression is restricted to information from the current frame only. Sure IBP is difficult to edit but compression wise it is superior to all-intra.
Click to expand...

IBP encoding falls apart when there is little in common between each successive frame, so the IBP encoder won't show any benefit over an All-I encoded video stream.
 
TheDingo said:
IBP encoding falls apart when there is little in common between each successive frame,
Click to expand...
Why would it fall apart?

The encoder would simply make it an i-frame or assign more bits to such cases just like it would assign less to simple cases.

This whole notion that IPB encoding is somehow a bad compression is just not warranted in my opinion.
 
Cary Knoop said:
This whole notion that IPB encoding is somehow a bad compression is just not warranted in my opinion.
Click to expand...

Each has their place, but IBP does not replace ALL-I encoding.

It falls apart when it either doesn't recognize that there is nothing in common between progressive frames so it tries to use B and P frames to encode the image sequence or when it simply runs out of bits to store image information.

IBP at 150 Mbps is not going to equal ALL-I at 400 Mbps when there is next to nothing in common between frames, the best case scenario here is IPB at 150 Mbps encoding with I frames versus All-I at 400 Mbps encoding with I frames, and in this specific case ALL-I at 400 Mbps wins.
 
Cary Knoop said:
That would be very untypical footage for anyone to watch. :)
Click to expand...

Which is why it is so difficult to compare these days. Back in the days of mpeg2 it was much more obvious to see the difference between all-i and IPB. I used to do a lot of 200-300mbps all-i mpeg2 HD video as a sort of competitor to ProRes and other similar formats. The Matrox I/O cards used that format and so did the Avid Liquid products. We also had a bunch of live video streaming boxes that could use up to 300mbps mpeg2. Pretty darn sweet format if you didn't mind 8bit.

The higher the bitrate the less likely you are to have macro blocking.

You are absolutely correct that IPB is more efficient. Thats its entire reason for existence. Once you reach a certain bitrate threshold however you no longer have much benefit over all-i. Its a similar concept for VBR DVD's with multi-pass encoding. Really good encoders that used multi-pass encoding could look at the frames and figure out visually how to replicate the same quality as the full 9.8mbps CBR encoding. So you could get away with 7, 6 or sometimes even 5mbps with the same visual look. At a certain point a group of frames no longer has enough detail to make more use of the extra bits. 400mbps all-i is already very high and we have seen just how clean it is on the GH5. 400 mbps IPB wouldn't really improve much because honestly what is there left to improve? At that point the IPB is no longer efficient either and much like all-i you are just wasting bits to pad each frame if it needs it or not. Chances are each frame doesn't need that much either. IPB makes sense when you want to target a smaller size and still yield great quality. Finding that sweet spot is the trick as it was with DVD encoding. Thats why we had multiple pass encoding.
 
Last edited:
Thomas Smet said:
That is again not very complex shots that would typically break an encoder. Especially one encoding HD IPB at 100mbps which is a ton of bits for HD. You need a lot more complexity in the shots to see a difference. If this is typical of the type of stuff one shoots then without a doubt IPB can be just as good.

I think there is a significant disconnect when it comes to understanding what can break an IPB encoding at a given bitrate. The higher the IPB bitrate the more it will retain a solid looking frame with no macro blocking because there are more than enough bits to handle the entire frame. Thats why using 400mbps for IPB is kind of silly because after a certain threshold you end up with frames that are basically the same as 400mbps all-i. They are just calculated and built in an inefficient way. But visually you pretty much end up with the same thing. The entire benefit of IPB is to have a lower bitrate while yielding as close to the all-i quality as possible in most cases. It is those rare cases (exploding cars/buildings, tree branches/leaves blowing in a wind storm, tires kicking up a spray of dirt/mud, a bubbling brook/stream) when the "lower" bitrate IPB has to make sacrifices and duplicate blocks between frames. If the scene is not very complex or if there are a ton of bits to share between the frames then visually the IPB and all-i should look the same or close to the same. It is only when the bitrate is too low (10-25mbps for HD, mpeg2) that each frame no longer has enough bits to visually reproduce each frame without some crude duplicating of blocks of pixels between frames.
Click to expand...
+1

Vesku's footage is well-shot, but it's almost totally lacking in the kind of busy details that would over-stress an H.264 encoder. Panasonic Lumix cameras have adaptive encoders that will only allocate as much bitrate as needed to encode each frame of a video at the encoder's targeted image quality. If your footage isn't pushing hard against the encoder's max bitrate limits, you won't likely see significant image quality issues.
 
I think the consensus is gathering momentum that the 400mbps codec is only of help in the editing stage rather than for any image quality gains. I feel also that in any codec the GH5 has certain artefacts when capturing motion which may not be made obvious until we see a GH6 with a possible global shutter or someone does some side by side tests with a high end cinema camera. My workstation doesn't struggle with the 150mbps codec but it's unusable on my current laptop so I may just continue with 400mbps for that alone until I get a laptop with an 8th Generation intel i7. The bottom line though is that 4k 10bit 4.2.2 is an amazing thing to have as an internal codec no matter what the codec is!
 
Lpowell said:
+1

Vesku's footage is well-shot, but it's almost totally lacking in the kind of busy details that would over-stress an H.264 encoder. Panasonic Lumix cameras have adaptive encoders that will only allocate as much bitrate as needed to encode each frame of a video at the encoder's targeted image quality. If your footage isn't pushing hard against the encoder's max bitrate limits, you won't likely see significant image quality issues.
Click to expand...

It was Mark Williams video.

GH5 ipb codec is variable but in practise it is quite constant and it is not dropping the bitrate in low detail scene. The fine gradations are detail too and needs lots of bitrate. GH5 ipb gives enough bitrate for low contrast parts and does not smooth then out.
 
Shirozina said:
I think the consensus is gathering momentum that the 400mbps codec is only of help in the editing stage rather than for any image quality gains. I feel also that in any codec the GH5 has certain artefacts when capturing motion which may not be made obvious until we see a GH6 with a possible global shutter or someone does some side by side tests with a high end cinema camera. My workstation doesn't struggle with the 150mbps codec but it's unusable on my current laptop so I may just continue with 400mbps for that alone until I get a laptop with an 8th Generation intel i7. The bottom line though is that 4k 10bit 4.2.2 is an amazing thing to have as an internal codec no matter what the codec is!
Click to expand...

Really depends on what you shoot. If you shoot a lot of complex stuff then all-i could have a quality benefit. Unfortunately 99.9% of tests online will not show you comparisons of that type of material and will typically involve a locked down camera shooting a still life scene with the author scratching their head why they don't see a difference.

If you don't shoot a lot of complex stuff then yes the 150 IPB is likely going to be exceptionally good and you may not gain much benefit over the all-i format.

These are not new concepts and many of us went through this with the GH1/2 hacks and the IPB vs all-i modes on the GH3/4. Sometimes in extreme sports or complex changes every frame type of stuff the all-i will be better. These are just choices at this point and honestly the 150mbps IPB and 400mbps all-i offer equal choices for different situations. Just another tool in the shed.

All-i exists strictly for peace of mind and consistency. When you shoot all-i you know every frame has an equal amount of data and built as a solid frame. There are no surprises. IPB looks fantastic and saves a ton of space and media costs but you could end up with some macro blocking every now and then depending on what you shoot. The all-i you don't really have to worry about it. Same thing when we use ProRes. It isn't that ProRes is vastly superior in terms of image quality but it means we get the same quality no matter how messed up each frame is. No surprises later on in post. To some professionals having that peace of mind that every frame is solid is worth the price of gold. To others it is a complete waste that 99.99% of people will never actually see.
 
Joe, Lee and others here have accurately described the benefits of All-I recording.

- Better motion/panning cadence thanks to avoiding any predictive errors
- Much better frame to frame imagery when it comes to large amounts of random motion (explosions, confetti cannons, etc)
- A lack of macroblocking errors thanks to each frame being a whole, instead of a predictive part of a previous frame

Personally, I'd shoot All-I for any narrative film stuff, as I want it to have as high a quality as possible for the edit and for the big screen, where all errors are amplified.

For the rest, I'll continue to shoot IPB as the GH5's codec really is very good in most instances.

Cheers,

Paul :)
 
You must log in or register to reply here.
Back
Top