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

V

Vardalos

Well-known member
Putting aside for the moment All-I as a benefit, my preliminary tests using an identical scene/exposure with a locked down GH5 indicate that it’s very hard to see any visual difference comparing 150 & 400Mbps clips on my 15” MacBook Pro Retina Screen.

The question is, if you can’t see a visual difference is there really a good reason to shoot at 400Mbps?
 
ozmorphasis said:
Easier on the computer when editing.
Click to expand...

Using Edit Ready, I transcoded both the 150Mbps and 400Mbps clips into Pro Res HQ @ 754Mbps and still cant see a visual difference. No problem editing on my late 2013 Quad Core MacBook Pro.

BTW, in Edit Ready and prior to transcoding the 4K 10bit clips playback great.
 
Vardalos said:
Using Edit Ready, I transcoded both the 150Mbps and 400Mbps clips into Pro Res HQ @ 754Mbps and still cant see a visual difference. No problem editing on my late 2013 Quad Core MacBook Pro.

BTW, in Edit Ready and prior to transcoding the 4K 10bit clips playback great.
Click to expand...

Right, so if you don't notice a difference in the image quality and you are nowhere near maxing out your processing of the IPB files, then FOR YOU, maybe there is no good reason for shooting in All-i. However, it is less taxing on processing resources compared to IPB so each person has to decide if that's reason enough.

As far as image is concerned, it generally will show advantages with very busy motion (trees with leaves, ripples on water, lots of panning, etc). A locked off shot of fairly static objects shouldn't look different between the two.
 
Ozmorphasis,

I could see a difference when I compared a clip recorded @ 400Mbps in camera to a simultaneously recorded Pro Res HQ clip recorded @ 754Mbps using my PIX-E5 there’s clearly more contrast and more color in the Pro Res recording.

Your comment about "Very Busy Motion" is well taken and I'll make more tests with that in mind. Perhaps others will also have some comments on this subject.
 
Vardalos said:
I could see a difference when I compared a clip recorded @ 400Mbps in camera to a simultaneously recorded Pro Res HQ clip recorded @ 754Mbps using my PIX-E5 there’s clearly more contrast and more color in the Pro Res recording.
Click to expand...
If it is "clearly" it probably means that levels or other things are not identical for both recordings. The differences should be virtually undetectable during normal viewing of frames and would only surface after more extreme grading.

Could you upload a comparison so we can see what you encountered?
 
Yeah, locked down isn't going to do anything to show the differences as even the LongGOP codec is going to be very simple for the camera to encode at good quality.

Throw some pans, moving objects etc at the scene, then you'll see the difference, especially in the way that areas of smooth colour don't block up as badly in All-I compared to LongGOP.
 
Great comments everybody, I guess my takeaway on all this is if you want the best possible 10bit 4K images out of your GH5 you’ll need an external recorder for very high Bit Rate recording. The Data Rate for Pro Res 422 HQ 4K is 754Mbps.

When an external recorder is not Practical but the subject matter does have “Very Busy Motion” shoot All-I @ 400Mbps.

If your subject matter is an interview or something static, you’ll get good results shooting LongGOP @ 150Mbps.
 
Vardalos said:
I guess my takeaway on all this is if you want the best possible 10-bit 4K images out of your GH5 you’ll need an external recorder for very high bitrate recording.
Click to expand...

In theory this should be true, but I would like to see some real-world examples where I can clearly see a difference between ALL-I 400 Mbps and an external recorder, because I have big doubts about this right now. ( the footage should contain both fast and complex motion, with few locked-down shots )

Personally, I would use an external recorder for other reasons than better than 400 Mbps image quality: HDR monitoring, daylight monitoring, low-cost media compared to V90 cards, etc...
 
Vardalos said:
If your subject matter is an interview or something static, you’ll get good results shooting LongGOP @ 150Mbps.
Click to expand...
You get good results in almost all scenes, they do not have to be static.

For instance pans are not going to be difficult for long GOP encoding as it is straight forward motion compensation. The velocity of the movements does not matter as well.

The only scenes that are going to be ineffective for long GOP encoding are scenes like running water, smoke and fire as there will be too much motion in different directions to effectively inter-frame compress. However variable bitrates will take the complexity of the motion somewhat into account by allocating more bits to hard scenes.
 
Cary Knoop said:
For instance pans are not going to be difficult for long GOP encoding as it is straight forward motion compensation.
Click to expand...

There's been a fair bit of discussion about GH5 "judder" when panning, that does not show up when shooting with higher end cine cameras. So it's possible this new 4K ALL-I format might fix this.
 
Long GOP is pretty advanced these days and it will be difficult to get the codec to really fall down compared to the All-I version imho. It is more about the editing and file size tradeoffs. Personally, I prefer long GOP codecs as it lowers your file size for card space and post backups while delivering great image quality. Kind of the modern best of both worlds for a lot of cases. But this is related to the type of work you do.

Would be interesting to see if the judder issue is related to the GOP. If it is then one strike against long GOP!
 
Cary Knoop said:
You get good results in almost all scenes, they do not have to be static.

For instance pans are not going to be difficult for long GOP encoding as it is straight forward motion compensation. The velocity of the movements does not matter as well.

The only scenes that are going to be ineffective for long GOP encoding are scenes like running water, smoke and fire as there will be too much motion in different directions to effectively inter-frame compress. However variable bitrates will take the complexity of the motion somewhat into account by allocating more bits to hard scenes.
Click to expand...

Are you suggesting that for almost everything 10bit you would go with 150Mbps, if you are, I suspect your right.

Admittedly my tests have all been locked down but I defiantly can’t see the difference between 10bit clips recorded at 150Mbps and 400Mbps there’s only an obvious difference when I compare a simultaneously recorded Pro Res Clip at 754Mbps. Then I do see a little more color and contrast in the Pro Res Clip.

BTW if you want judder free pans switch to 30fps.
 
Hi

Vardalos said:
Are you suggesting that for almost everything 10bit you would go with 150Mbps, if you are, I suspect your right.


Admittedly my tests have all been locked down but I defiantly can’t see the difference between 10bit clips recorded at 150Mbps and 400Mbps there’s only an obvious difference when I compare a simultaneously recorded Pro Res Clip at 754Mbps. Then I do see a little more color and contrast in the Pro Res Clip.

BTW if you want judder free pans switch to 30fps.
Click to expand...

There is a certain irony that people argue motion is better with ALL-I, however we generally, certainly for the case of 24fps, ensure motion is blurred in order to avoid judder due to the slow frame rate, so there is no extra detail really for ALL-I to capture.

Also, take 24fps, with ALL-I that means each 4K frame is compressed to approx 2MBytes regardless, yet for IBP, it only needs to record the changes, so you could have the first I frame in a second of video compressed using 50% of it's data rate, so 75Mbits/sec, giving the frame a compression size of 9Mbytes, then using the remaining to encode just the changes, at the end of the GOP in that 1 second, is that last frame going to look worse or better than a frame compressed to 2MBytes? This is probably why people aren't seeing much difference. Also long GOP with a complicated scene could add several I frames in that one second with those I frames still being assigned more bits than each frame of ALL-Intra at 400Mbits/sec.

Basically ALL-I is wasting data on encoding frames entirely for the benefit of easier editing, whereas IBP uses the redundancy to maintain detail. Long GOP in 4K at lower bit rates than the 150Mbits/sec still look pretty good, so I don't think the GH5 is struggling with IBP at 150Mbits/sec.

The 400Mbits/sec on the GH5 is more about marketing and competing and because they can on the GH5, it's the usual bigger numbers sell type of thing, rather than an obvious in your face improvement.

As for colours or contrast looking better on one or the other, that isn't a difference caused by ALL-Intra versas GOP, that isn't how it works.

Regards

Phil
 
Vardalos said:
Are you suggesting that for almost everything 10bit you would go with 150Mbps, if you are, I suspect your right.

Admittedly my tests have all been locked down but I defiantly can’t see the difference between 10bit clips recorded at 150Mbps and 400Mbps there’s only an obvious difference when I compare a simultaneously recorded Pro Res Clip at 754Mbps. Then I do see a little more color and contrast in the Pro Res Clip.

BTW if you want judder free pans switch to 30fps.
Click to expand...

You see more color and contrast in the external ProRes because of how most external recorders handle the 0-255 luma range. There is a slight shift in contrast due to this difference since most external monitors/recorders expect a 16-235 luma range video. The Convergent Design Odyssey is the only monitor/recorder I know of that can compensate for this difference in luma values. The difference you see really has nothing to do with the 754mbps or that it is ProRes and is just the luma difference.

If shooting in a profile that is not v-log and using an external recorder/monitor I would suggest setting the GH5 to 16-235 range.
 
Thomas Smet said:
If shooting in a profile that is not v-log and using an external recorder/monitor I would suggest setting the GH5 to 16-235 range.
Click to expand...

Maths is not my strong point, but since the discussion seems to be about 10 bit wouldn’t the Legal range be 64-940?

And what is the recommended Luminance Level for VLog-L?
 
Bern Caughey said:
Maths is not my strong point, but since the discussion seems to be about 10 bit wouldn’t the Legal range be 64-940?

And what is the recommended Luminance Level for VLog-L?
Click to expand...

yes it is. I tend to always write it in terms of the 8bit values for simplicity reasons. I also use a GH4 which labels the luma as 0-255 or 16-235 even though the HDMI output is 10bit.
 
I'm reading a lot of the same misconceptions here regarding all-i vs long-GOP that seem to never go away.

- You can't directly compare bit rates between the two compression methods as a way to ascertain picture quality. That's a fruitless exercise.

- You can't compare the bit rates of two different codecs (ProRes & h264) as a way to ascertain picture quality. For example, h265 is roughly twice as efficient as h264, requiring half the bandwidth to yield the same PQ.

- Inter-frame compression is the same between both flavors of h264. It is only intra-frame compression that changes. A static scene will look the same...it's the same codec...assuming adequate available bandwidth.

- You use the all-i flavor for 2 possible reasons (of course there are more, but 2 main reasons):
1. It requires less processing power to decompress. This could allow a system that otherwise could not playback in realtime to do so, or allow systems that can to do more post-processing (filters) in realtime.
2. It removes the chance of synthetic motion artifacts. Long-i codecs predict motion, and sometime predict incorrectly in ways that are visible. (On the GH-series, such an artifact can be induced by panning, for example.)

- You use the long-GOP favor for one main reason: It uses less bandwidth/storage space.

tldr;
Assuming that 400Mbps is adequate to resolve your scene (which it probably is) and you have big enough/fast enough cards and hard drives then there is no good reason to *not* shoot all-i.
If you don't have/don't want/can't afford the necessary bandwidth and storage requirements, shoot long-GOP.
 
You must log in or register to reply here.
Back
Top