Codecs Clarified, Please?

My inquiries regard Resolve 18, shooting video on the BMPCC4K and Panasonic GH5 with or without the Ninja V. I basically grasp the differences between ALL-I compression (AVC-Intra) and Long GOP (I pretty much stay away from all Long GOP if I can help it). Here is where my confusion sets in and I just can't seem to find the answers combing the internet profusely.

1) I understand the benefits of using Prores 422 10 bit HQ internally with the BMPCC4K and with the Ninja V attached to my GH5 as .mov files.
2) The Panasonic GH5 shoots, internally, 422 10 bit ALL-I (intraframe) at 400 mb/s also using a .mov wrapper, however, its codec is h.264. Therein lies my confusion with this comparison between Prores and .h264 codecs. In this case, both are Intraframe and both use high bitrates.

Will someone kindly explain the compression differences between the two? Prores in the .mov wrapper vs h.264 in the .mov wrapper using Panasonic's internal, Intraframe, high bitrates? I know .h264 is usually a finishing codec (for DVDs or so I've read) but because the GH5's internal h.264 is Intraframe with a much higher bitrate, does that cancel out the fact that it's compressed? Or is it a different kind of compression than regular h.264? I ask all of this to ascertain if it will make one bit of difference to shoot and edit more with Panasonic's internal codec or the Prores through the Ninja V? How does the difference affect editing with Davinci Resolve 18?

I hope I've articulated this inquiry clearly enough.

Regards,
Hugh

Different compression technologies have different efficiencies, therefore H265 is better, or more efficient, than H.264, for example, since it is newer and more advanced. H.264 is not usually a finishing codec, nor in DVDs, which are Mpeg, though Blu-Ray can be H264. ProRes, as is DNxHD/R, is a traditional post codec, which is relatively light on CPU and very efficient in processing terms, as opposed to H.264, which requires rather more CPU decoding. It has become a capture camera codec too since the workflow and path to post, make it relatively painless and the compression is perceptually lossless. The downside being it is much larger than more efficiently compressed schemes, like H.264/5. Given sufficient bandwidth, bit depth and subsampling any codec can produce the required quality necessary. So in broad answer to your question, as long as the fidelity and quality of recording is sufficient and your industry workflow, processing and tools can handle it and accept it, it is immaterial which codec you use.

Thank you for taking the time to explain. I'll probably just capture on the Ninja V when using the GH5. So, to be clear: even though my GH5 internal codec is .h264 400 mb/s ALL-I, it's still more compressed than Prores HQ and therefore more taxing on my CPU?

So, to be clear: even though my GH5 internal codec is .h264 400 mb/s ALL-I, it's still more compressed than Prores HQ and therefore more taxing on my CPU?

Yes because there is a confusion between mb/s and mbps. The 400 you speak of is mbps not mb/s. Prores HQ @ UHD is 220 mb/s, which is 1760 mbps. So you can see 400 mbps is significantly more compressed than Prores HQ. On the Ursa Broadcast G2, for example H.265 at the same resolution (and 4:2:2 10 bit, as well) is only 24.6 mb/s, or around 200 mbps. So again, 400 mbps is around 50 mb/s.

My good man, have you any idea how long I've been waiting for that clarification? Thank you for explaining in clear, concise terms that even a non-technical guy like myself can understand.

I just found this:

https://atomos.zendesk.com/hc/en-us/art ... Data-Rates

I shoot UHD ALL-I on my GH5 using my Ninja V to capture to Prores HQ which the above chart says captures at 737 MB/s. Converted to mbps that's a whopping 5656! Is that possible?

Yes, it's possible but those are completely different figures to Black Magic. I took 220 mb/s 'max data rate' direct from the Ursa Broadcast G2 manual. However given that DNxHD @ 1080 can be 185x/220x (very similar to Prorez HQ), then UHD could easily be 4 x that data rate. It doesn't explain why BMD and Atomos have such wildly different figures though. I am of the opinion that Atomos are confusing mbps with mb/s too, because as you can see from the chart 1768 'mb/s' for 2160p60 tallies with BMD's max data rate for the same codec, at the converted mbps I quoted. Since I also know that the DNxHD rates I quoted actually work out at 20 something mb/s and could be easily captured to a standard 7200rpm usb 2 attached drive, that is only capable of a little more throughput themselves.

I would film the same complex scene (to challenge the encoders) both internally on the GH5 in H.264, and externally on the Ninja in the various ProRes flavors.

Then pixel peep to compare, and choose the right option for your needs.

Thank you, gentlemen. It irks me that two different products can quote different specs.

I just discovered yet ANOTHER online discrepancy between the GH5 ALL-I h.264 400 Mbps codec and the BMPCC4K Prores HQ 220Mbps claims. This is sort of becoming annoying. There seems to be confusion about terms:

Mbps = Megabits per second. MBps = Megabytes per second. Megabytes are eight times larger than megabits. I get that. Panasonic claims its h.264 ALL-I UHD codec is 400 Mbps. I read Apple ProRes 422 HQ is 220Mbps. Why are people saying the ProRes files are larger when Panasonic claims to have 400 Mbps vs the ProRes 220 Mbps? Both are intraframe. I believe a gentleman on this thread advised ProRes is 220 MBps and not Mbps but an online article claimed ProRes 422 HQ is 220 Mbps. May I get some clarification? Could the Panasonic codec be smaller because h.264 is more compressed than Prores? Do the Mbps speeds matter in this case?

Thank you kindly!

I just discovered yet ANOTHER online discrepancy between the GH5 ALL-I h.264 400 Mbps codec and the BMPCC4K Prores HQ 220Mbps claims.

You’re comparing HD Prores to 4K h.264.

Prores HQ is 880 Mbps @ 4K.

Good Luck

You’re comparing HD Prores to 4K h.264.

Prores HQ is 880 Mbps @ 4K.

Yes Hugh, please re-read the thread. Mb/s = megabytes per second; mbps = megabits per second. Prores Hq = 220 mb/s @ 2160p60 or 1760/8 mbps. Your H.264 is 400 megabits per second, therefore it is considerably more compressed than Prores HQ. It is Atomos and not BMD, who are in error to quote MB/s, when they meant and should have proofed mbps, which admittedly is not helpful for you. Sometimes too megabytes per second can be expressed confusingly as MBps rather than mb/s, so watch out for that too.

You’re comparing HD Prores to 4K h.264.

Prores HQ is 880 Mbps @ 4K.

Yes Hugh, please re-read the thread. Mb/s = megabytes per second; mbps = megabits per second. Prores Hq = 220 mb/s @ 2160p60 or 1760/8 mbps. Your H.264 is 400 megabits per second, therefore it is considerably more compressed than Prores HQ. It is Atomos and not BMD, who are in error to quote MB/s, when they meant and should have proofed mbps, which admittedly is not helpful for you. Sometimes too megabytes per second can be expressed confusingly as MBps rather than mb/s, so watch out for that too.

they will never understand.

i read this so often, because the marketing departments
of sony, pana and canon seem to do a good job;- )

.

Finally. Thank you. I thought I was losing it. My confusion stems from others' marketing inaccuracies. This is exacerbated by the fact I gave up caffeine.

FYI: This may or may not be relevant to the compression question on this thread, but, I was on the fence about shooting content with my internal Panasonic GH5s h.264 ALL-I .mov 422 10 bit codec vs using an Atomos Ninja V and capturing Apple Prores 422 HQ. I am no longer on the fence and I have now made my decision to never use the camera's internal codec again. Why? Whilst attempting to cut and paste a four minute HD clip shot with the camera's internal ALL-I codec from the SD card (via reader) to my editing computer hard drive, the transfer froze up in the middle of the transfer. I made three more attempts and it kept freezing. The only way to transfer was to copy and paste, not cut and paste. Immediately after, I shot much longer video clips on the same camera with a Ninja V and the ProRes files cut and pasted from the SSD was smooth, flawless and extremely fast. Despite the fact tha camera's internal codec is ALL-I, 10 bit 422, it's still h.264, very compressed. I also compared side by side footage of the internal codec vs Apple ProRes from the Ninja and the internal codec did not hold up as well as the Ninja codec to grading as evidenced by a blowup. The Apple codec seemed to also have more dynamic range. Now I don't mind the little bit of extra inconvenience using the Ninja V on my run and gun documentary assignments. The Apple ProRes files are just better.

This "prores is the best thing since sliced bread and H.264 and H.265 is crap" myth seems to perpetuate the internet.

For bitrates over 400Mb/s all CODECs start to converge in terms of quality.

For lower bitrates and comparing the same bitrate and all-intra compression:

1. H.265 is the best, it uses the latest advanced compression methods.
2. H.264 is second and works well for HD and lower resolutions.
3. Prores is last, the compression technology is older and less efficient. Quality degrades rapidly with bitrates going under 250Mb/s

Howard Roll is correct in stating the Prores HQ @4K is 800mb/s

The notion that Mb/s and Mbps is different is nonsense. A small case b means bits while an upper case B means byte.

Howard Roll is correct in stating the Prores HQ @4K is 800mb/s

The notion that Mb/s and Mbps is different is nonsense. A small case b means bits while an upper case B means byte.

1768 was taken from the BMD manual for UHD 60, which I specifically stated; obviously it varies according to framerate. The difference between megabytes and megabits is very real and not nonsense; which is why Hugh was understandably and forgivably confused in the first place. Who the hell cares about whether it's a capital or not?

The difference between megabytes and megabits is very real and not nonsense;

It most certainly is not nonsense.

Who the hell cares about whether it's a capital or not?

I would say anybody who respects the convention that b = bits and B = bytes and realizes that the difference between megabytes and megabits is very real and not nonsense

And Mb/s and Mbps mean the same thing, however, Mb/s and MB/s are quite different!