ProRes VS H.264

Hello,

I've currently been filming everything on a Canon 7D. The 7D records approximately 48mb/s H.264, and I was wondering what ProRes flavour (on the URSA Mini 4.6K) would be approximately equal quality to the 7D? I tried to find information all over the web, though the best I could find was "Prores VS H.264" for editing. I realise that ProRes is better for editing purposes.

Thanks in advance.

Equal to the 7D's H.264 recording? Probably just ProRes Proxy or maybe ProRes LT. At 1080p, ProRes Proxy runs at 36bps and LT runs at 82Mbps.

Note that ProRes Proxy is the lowest quality video format the URSA will record.

URSA will also record in ProRes LT, ProRes 422, ProRes HQ, ProRes 4444 and CinemaDNG/RAW. RAW of course is the highest quality.

For reference, ProRes 4444 at 1080p is about 264 Mbps.

This begs the question though - why but the 4.6k URSA just to get it to record like a 7D when you already have a 7D?

Don't worry xD. I most likely won't be filming in Proxy, it's just good to know that I can use it, and expect the same video quality as the 7D's. It's also good to know, if I do need that extra storage space, that I am able to get the same video quality as the 7D using the URSA Mini's worst codec. I definitely won't be buying multiple Cfast 2.0 cards, just going to be one 128GB one. So if I tried to record long events in 422HQ, or Raw, I would end up out of storage space. Though I most likely would still be filming Raw 4:1 (4K), for anything where storage isn't a problem.

Prores Proxy surely does NOT feature the same IQ of your 7D.
Prores Proxy already blocks the image a lot, so you can see it even without pixelpeeping. As Prores is an All-I Codec in any flavour, you obviously need a higher Bitrate then the h264 of a 7D.
Prores Proxy is around 36mbit, so even lower than your 7D; use AT LEAST LT.

However, from the overall image aesthetics and "quality", it's not really possible to drag a BMD down to a 7D only with a different Prores flavour

Yeah I'd stay away from prores proxy. It's pretty gross looking, especially in low light. I find when I need longer record times LT is fine. My target is YouTube, so there's often no sense in going much higher.

My gh3 records in h264 and it looks way better than proxy on my BMPCC. I can't remember off hand but I think the bit rate is a bit higher than pro res proxy runs.

ProRes is a far superior codec than H.264, this is long time known. The quality and color rendition is ways better, you'll notice it at first sight. I had worked along some folks that try to combine Canon T5i, 7D and 5D Mark III with the original BMCC 2.5K set to 1080p and the color grading is tricky, even when setting the cameras to similar ASA, f-stop, and white balance, pretty tricky I'd say, the sharpness and quality of ProRes is instantly noticeable, as I've already stated.

Dylan Neild already gave you fair enough info about the the data ProRes can handle. A 7D won't ever get the same native video quality than a BMCC or URSA at 1080p, ProRes Proxy has too much artifacts and low readout, so your only options are ProRes LT and ProRes 422, which are still superior than H.264.

However, I don't know if image quality will improve if you work with an external recorder as Atomos or Odyssey plugged to your 7D to make it shoot ProRes or even higher resolutions than a DSLR is capable of.

Prores proxy is 10 bit. H264 8 bit. Fhd is 2mpix, still cameras are 18-22-36mpix and more.
Alaising, lack of colour information, low detailed shots - that's h264 from a stills camera. C100-c500 with h264 are better, but every prores is beautifull. Even proxy and 25 hours of shooting fhd via 480gig Ssd.
So:
1) U won't be able to reach low quality of 7d out of the ursa mini. Only after decoding to 8-bit, bluring and so on.
2) U will be able to shoot a lot in fhd prores proxy and prores LT with better than 7d's h264 quality
3) If there's a problem of 2nd cam and U wanna use 7d - use magic lantern for 7d or buy micro cinema camera, pocket cinema camera and use prores. Postprocessing ursa mini footage to reach a 7d look is useless, as I think.

Prores proxy is 10 bit. H264 8 bit. Fhd is 2mpix, still cameras are 18-22-36mpix and more.
Alaising, lack of colour information, low detailed shots - that's h264 from a stills camera. C100-c500 with h264 are better, but every prores is beautifull. Even proxy and 25 hours of shooting fhd via 480gig Ssd.
So:
1) U won't be able to reach low quality of 7d out of the ursa mini. Only after decoding to 8-bit, bluring and so on.
2) U will be able to shoot a lot in fhd prores proxy and prores LT with better than 7d's h264 quality
3) If there's a problem of 2nd cam and U wanna use 7d - use magic lantern for 7d or buy micro cinema camera, pocket cinema camera and use prores. Postprocessing ursa mini footage to reach a 7d look is useless, as I think.

I was going to say that.

You also need to remember that ProRes Proxy records at 10-bits (from what I've read), so it's storing 4x more colour data than H.264, which is why a lot of blocking and banding appears as there's less data for other finer details.

Sorry guys, but this thread is full of technical inaccuracies.

Fact 1)
10bit encoding saves bandwidth above 8bit encoding.
Why? See here: Why does 10-bit save bandwidth (even when content is 8-bit)?
And here: Quciktime 16-bit

Fact 2)
ProRes vs h.264

ProRes basically uses an all-I flavor of MPEG-2 encoding - that's the reason why you see all the blocking with highly compressed ProRes variants like Proxy or LT. Proxy @ 1080p25 has 38 Mbit/s, and LT has 85 Mbit/s. Both use 4:2:2 chroma subsampling and 10bit.

h.264 is a very efficient encoding scheme - far better (around 3 times more efficient) than MPEG-2. But to be that efficient it has to take advantage of GOPs and a lot of encoding tricks. It has much better motion compensation, can use more B-frames much more flexible, much less ringing around sharp contrast edges. It supports chroma subsampling up to 4:4:4. And bit depth at 8, 10, 12, and 14-bit per sample.
Of course encoding in camera hardware will not make use of all tricks.

If 4:4:4 = 100% (no chroma subsampling), then 4:2:2 = 66,67%, and 4:2:0 = 50%.

So the 4:2:0 of camera h.264 codec already has less data to compress than ProRes with its 4:2:2.
24 Mbit AVCHD equals ~ 72 Mbit MPEG-2 + 16,67% = 84 MBit/s

So one could say ProRes LT is kind of a match to AVCHD, ProRes Proxy is definitely not.
In realty even ProRes LT is no match, watch this eye-opening video here:

Fact 3)
10bit vs. 8 bit is highly overrated. If you do color correcting, it is much more important that your software works in high bit depth like 10bit/16bit integer or 32bit float like AFX and Resolve do.

See here: Visual Quality Difference Between 10-bit and 8-bit precision in post processing

And if you really believe 10bit source footage makes such a huge difference during grading - prove it.
Take your DNG and render it once to 10-16 bit and once to 8 bit. Then do some heavy CC in 32bit mode.
Show your results.

Sure it makes a huge difference. Film some fog, then pull up the highlights some more and darken the shadows a bit... Same for sky gradients with or without clouds. Contrast increases too.
You get 4 times the number of steps and thus the gradient falls apart much later.

Gesendet von meinem A0001 mit Tapatalk

To be quite honest I shoot Proxy for long shoots. like a 3-4hr concert. No problem. If your shooting a film or commercial i wouldn't recommend it but this is for Public access Television. especially since they don't have the room to store all that footage.

Sorry guys, but this thread is full of technical inaccuracies.

Fact 1)
10bit encoding saves bandwidth above 8bit encoding.
Why? See here: Why does 10-bit save bandwidth (even when content is 8-bit)?
And here: Quciktime 16-bit

Fact 2)
ProRes vs h.264

ProRes basically uses an all-I flavor of MPEG-2 encoding - that's the reason why you see all the blocking with highly compressed ProRes variants like Proxy or LT. Proxy @ 1080p25 has 38 Mbit/s, and LT has 85 Mbit/s. Both use 4:2:2 chroma subsampling and 10bit.

h.264 is a very efficient encoding scheme - far better (around 3 times more efficient) than MPEG-2. But to be that efficient it has to take advantage of GOPs and a lot of encoding tricks. It has much better motion compensation, can use more B-frames much more flexible, much less ringing around sharp contrast edges. It supports chroma subsampling up to 4:4:4. And bit depth at 8, 10, 12, and 14-bit per sample.
Of course encoding in camera hardware will not make use of all tricks.

If 4:4:4 = 100% (no chroma subsampling), then 4:2:2 = 66,67%, and 4:2:0 = 50%.

So the 4:2:0 of camera h.264 codec already has less data to compress than ProRes with its 4:2:2.
24 Mbit AVCHD equals ~ 72 Mbit MPEG-2 + 16,67% = 84 MBit/s

So one could say ProRes LT is kind of a match to AVCHD, ProRes Proxy is definitely not.
In realty even ProRes LT is no match, watch this eye-opening video here:

Fact 3)
10bit vs. 8 bit is highly overrated. If you do color correcting, it is much more important that your software works in high bit depth like 10bit/16bit integer or 32bit float like AFX and Resolve do.

See here: Visual Quality Difference Between 10-bit and 8-bit precision in post processing

And if you really believe 10bit source footage makes such a huge difference during grading - prove it.
Take your DNG and render it once to 10-16 bit and once to 8 bit. Then do some heavy CC in 32bit mode.
Show your results.

I'm not so keen in it like you, far far not like you. But I know that in the question asked my answer is true. If h264 overally is cool - ok, let it be so. 7d's h264 is not.
I'm not going to say that prores is an industry standart and so on, cause that's the only my argument.
I just ask you not to hack brains of those, who ask simple questions like that. I'm sure, h264 won't suffer from our words that way more better to shoot prores. Even proxy.

I'm not so keen in it like you, far far not like you. But I know that in the question asked my answer is true. If h264 overally is cool - ok, let it be so. 7d's h264 is not.

h.264 is a very, very good codec. Every well done Blu-ray shows that - especially since the x.264 encoder matured which can even preserve fine details like grain/noise.

The Canon C100 provides proof that a camera can make fantastic AVCHD recordings when everything else is high quality. The 7Ds problems are not the codec - it is the way how the camera downsamples the sensor's image and how advanced the internal hardware encoder is. I am almost sure the C100 uses a much better implemented hardware encoder than the 7D.

I'm not going to say that prores is an industry standart and so on, cause that's the only my argument.
I just ask you not to hack brains of those, who ask simple questions like that. I'm sure, h264 won't suffer from our words that way more better to shoot prores. Even proxy.

The problem with ProRes is its age. It shows. And Apple does not improve their ProRes decoder for Windows, where 4k ProRes is VERY cpu heavy to play - like 80-90% on a 6 core i7.

I really would prefer a better codec like Cineform - in 4k playback it utilizes just around 15% cpu. And it delivers much better quality while using less bandwidth. For example there is no mosquito noise around sharp edges and no visible blocking at reasonable bitrates.
Unfortunately Cineform was bought by GoPro and it seems development for it has stopped. But it has been chosen as an open SMTPE standard now. So maybe this will lead to cameras recording in Cineform in the future.

Things are simple- ProRes is an intermediate codec, where long GOP h264 is rather for final delivery (web, Blu-ray etc). One was developed to slightly reduce bandwidth mainly for high resolution videos, where other for providing maximum quality at lowest possible size. One is relatively low on CPU, other is very intensive (try decoding UHD 60p AVC file purely in CPU mode).
ProRes itself is quite plain and comparable to something like MJPEG, where h264 is way more complex.

What you can do is to compare ProRes to AVC-I or XAVC (which is AVC-I for HD+). This will be fair comparison- similar bitrates and I frame nature.
As mentioned- well done Blu-ray can look very good against ProRes master which is was made from (not taking problem with gradients into account), but there is nothing surprising here. Compare 35Mbit AVC-I against 175M ProRes and suddenly it will look crap. There is also other factor here: Blu-rays are done with state of the art (assuming good releases), multi pass h264 encoder, where recording in cameras is quite often 'compromised' for realtime processing and energy efficiency. As mentioned there are also better and worse h264 hardware implementation in camera, so this can also make a difference, but long GOP will be always waaaay more effective than I frame only. Nothing to debate here about. Even h264 is not good enough to provide better quality for the same bitrate in I frame only mode, against MPEG2 long GOP.

Question: what ProRes mode is equivalent of 48Mbit h264 file from Canon is bit wrong. If you convert your Canon file to any ProRes file you always loose quality. It's up to you- if you want to preserve 99% quality than alway use at least ProRes HQ, which is good enough for most cases. Latest XQ mode is very good, because compression is low, about 4:1. 3:1 in reality can be treated as uncompressed and still saves 3x storage space and bandwidth! (Sony F65 RAW is about 3:1 compressed). I would even argue that there is no point to have uncompressed recording, as difference in reality (after whole post processing) will be basically none.

Mentioned Cineform in terms of technology is miles above ProRes, it's very unique. Like JPEG2000, but few times faster to encode and decode. It's almost symmetric, so encoding and decoding takes the same time. The best part of it is fact that if you have 4K file you can play it just at 1:2, or 1:4, etc. resolution and use waay less CPU for it (every current laptop will play Cineform 4K file). This is due to wavelet based technology and it's not possible with ProRes, DNxHD or GV HQX etc. It's like having proxy file, but not next to the main file, but inside it Another Cineform unique feature is support for RAW data- you can take e.g. DNG RAW data and convert to Cineform RAW (so you still have RAW data, but as Cineform compressed file). This can reduce size massively and keep great quality.
The only reason why it's not mainstream is fact that it was developed by 1 guy, not a big company, like Apple or AVID. Latest SMPTE work may help, but even more than this fact that Adobe products have native support for Cineform now. Cineform tends to use even lower bitrates than ProRes, but because of this sometime quality is worse (depending on the quality setting). One important thing during any quality comparison is the source. You have to use never DCT compressed file (straight from good camera). Many comparisons out there are done on files, which already went through ProRes or other DCT based compression and this will make ProRes looking way better. On a never compressed (or just compressed in camera before debayer file) Cineform will perform very well.

h264 is a very good codec, but even with todays fast CPUs use of AVC-I and XAVC is still relatively low, because this is still more demanding than ProRes, DNxHD or Cineform. h264 4K long GOP is another few levels more complex.

Problem with ProRes is not really its age- it still has no real competition, so why changing it? (beside Apple is not interested in PRO market anymore at all). PC slow decoding is more a QuickTime on Windows problem- it works well on Mac and also on PC/Linux in software with native ProRes implementation.
DNxHD/R is quite bad- newest but least efficient and last choice on my list.