4:2:0 vs. 4:2:2
- Thread starter D10S
- Start date
...Way back when the Crews.TV guys posted a downloadable ZIP file of uncompressed screen grabs comparing native AVCHD 4:2:0 footage against AVC-Intra 4:2:2 100 Mbit footage.
Crews.TV : AF100 Stress Test
Crews.TV : AF100 Stress Test
Did you think you were in the F3 forum when you posted? The AF is 8 bit not 10 bit and how is looking at the F3 footage going to help?
Personally I'd like to see some more full res frame grabs from shots with motion as there has been contradictory opinions on how much difference it makes, e.g. BBC Scotland saying "obvious softening on motion" with the on board codec.
Brian@202020
Veteran
You actually do see a difference. I shot some Nano 280 Mbps stuff and compared it to the native codec a while back and the 280 Mbps 422 stuff has a deeper, richer, more saturated feel to it. That said I use the Native codec 99% of the time because it's good 99% of the time for the stuff I do. I'll reserve the Nano for green screen and projects that need a more cinematic feel.
Benjamin Eckstein
Well-known member
I would love to see some graded comparisons too as that is probably the most useful. Ran into some serious blocking/banding today when I used some Magic Bullets (which I rarely) use on some AF native footage. It was a pretty deep look to it, but it certainly blocked out more than I expected. Definitely would have been interested to see it with Nano files too.
Barry_Green
Moderator
The Nano doesn't help with banding, it has the same 8-bit issues. I posted some files for someone here, Hywel I think, and he graded them both and said that a) he understood why so many people say that there isn't much of a difference, and b) the banding was the same in both.
Olof's comparison seems to show an improvement on banding here -
http://www.dvxuser.com/V6/showthrea...ot-with-Nano&p=2239992&viewfull=1#post2239992
But not in this one -
http://www.dvxuser.com/V6/showthrea...in-the-AF100&p=2231187&viewfull=1#post2231187
http://www.dvxuser.com/V6/showthrea...ot-with-Nano&p=2239992&viewfull=1#post2239992
But not in this one -
http://www.dvxuser.com/V6/showthrea...in-the-AF100&p=2231187&viewfull=1#post2231187
Brian@202020
Veteran
Like Barry said, banding is an 8-bit issue. That said my experience is the Nano does actually improve on the banding a little. This is because noise hides the banding, and the AVCHD codec hides the noise, hence more banding on the AVCHD footage. The SDI and HDMI output has more noise, but this noise is more pleasant to the eye than noise in the AVCHD codec which can become blocky. Bottom line is the way AVCHD 420 is set up; doing heavy grading means that noise is introduced which can become blocky and/or banding can become a problem. In my experience the AF100's AVCHD codec does hold up better than the XDCAM codec so light grading seems to be fine.
As Barry said, when I looked at the footage he'd kindly recorded for me with the onboard codec and the Nanoflash, I could not see much difference in the banding. This was a worst-case scenario for banding (a mannequin head against a plain white wall which was glancing lit, so the smoothest "real" gradient one is likely to encounter). It was not designed to stress the codec in any other way- I'm sure you'd start to notice the shortcomings of the on-camera codec a lot more if there was a lot of motion in the scene plus a fast camera move, for example.
For this specific "tranquil" test the Nanoflash made very little difference to the banding and we deduced that it was probably an 8-bit issue (with interplay with camera settings- it looks more natural with a bit of noise in the image to smear out the banding, for example).
That's not to say one wouldn't benefit from external recording at a higher bitrate for other scenes. Clearly if one is doing chromakey one's first request would be to want 4:2:2 as a starting point, and motion-heavy scenes would probably benefit a lot more from an i-frame only codec. But for the specific case in which I was experiencing banding, we concluded that an external recorder was not a panacea especially not at Nanoflash prices (the Ninja may be worth it purely as a dual-record/hardware ProRes transcoder leaving aside any benefits from 4:2:2 straight off the camera).
Following the advice of Shooter and others, I've found my best results to be to push the exposure up to the very top of the scale, rather than allowing myself grace at the 100% IRE level I push the whites to 105%. This is a bit like Russian roulette in that if you push a bit high over 107% and overexpose, stuff can clip in an ugly way. But it does generate much cleaner images and reduce the need for grading, and hence avoids over-exaggerating the banding.
Personally I am much happier with the AF100 shooting now, although it does feel a bit like walking a knife edge the results are much, much better. Frankly I'll be looking for a more "RAW-like" shooting style if/when funds allow (stills background = used to shooting RAW with 13+ stops of dynamic range and choosing the look in RAW processing).
I decided to save my money and not buy an external recorder just yet after seeing the footage Barry sent me as I'm not convinced of the benefits for the footage I'm currently shooting.
Cheers, Hywel
For this specific "tranquil" test the Nanoflash made very little difference to the banding and we deduced that it was probably an 8-bit issue (with interplay with camera settings- it looks more natural with a bit of noise in the image to smear out the banding, for example).
That's not to say one wouldn't benefit from external recording at a higher bitrate for other scenes. Clearly if one is doing chromakey one's first request would be to want 4:2:2 as a starting point, and motion-heavy scenes would probably benefit a lot more from an i-frame only codec. But for the specific case in which I was experiencing banding, we concluded that an external recorder was not a panacea especially not at Nanoflash prices (the Ninja may be worth it purely as a dual-record/hardware ProRes transcoder leaving aside any benefits from 4:2:2 straight off the camera).
Following the advice of Shooter and others, I've found my best results to be to push the exposure up to the very top of the scale, rather than allowing myself grace at the 100% IRE level I push the whites to 105%. This is a bit like Russian roulette in that if you push a bit high over 107% and overexpose, stuff can clip in an ugly way. But it does generate much cleaner images and reduce the need for grading, and hence avoids over-exaggerating the banding.
Personally I am much happier with the AF100 shooting now, although it does feel a bit like walking a knife edge the results are much, much better. Frankly I'll be looking for a more "RAW-like" shooting style if/when funds allow (stills background = used to shooting RAW with 13+ stops of dynamic range and choosing the look in RAW processing).
I decided to save my money and not buy an external recorder just yet after seeing the footage Barry sent me as I'm not convinced of the benefits for the footage I'm currently shooting.
Cheers, Hywel
P.S. This is straight out of the camera, ungraded, shot to get the highlights at 105% IRE.
You can see some banding on the background but it is much improved over what I was getting before when I was shooting to get the highlights at 100% IRE:
which admittedly is post a fairly significant grade but a lot of that was to get the overall exposure level right before I started mucking around!
Cheers, Hywel.
You can see some banding on the background but it is much improved over what I was getting before when I was shooting to get the highlights at 100% IRE:
which admittedly is post a fairly significant grade but a lot of that was to get the overall exposure level right before I started mucking around!
Cheers, Hywel.
Barry_Green
Moderator
I was using typical low-noise settings, so yes banding would have been more likely (in both). The more noise in the signal, the less banding will show up (in both). And Brian's observation is true that you can notice more noise in an external high-bitrate recording, because the AVCHD codec ignores some of the noise. And yes, I do agree that the AVCHD codec holds up better than XDCAM codecs so you might be able to squeeze a little more out of it in post. But you're not going to go yanking it up or down three stops' worth without seeing serious degradation, obviously.
Barry_Green
Moderator
Agreed absolutely, and one of the reasons I have a Nano.
For light-duty recording like headshots and interviews, I really pretty much defy anyone to see much of a noticeable difference, if at all, from an external recorder. If you're dealing with chroma keying you'll see a very big and very welcome improvement with the external recorder (as long as it's set to record 4:2:2; remember the Nano can be set to 18mbps or 35mbps mode and those are 4:2:0 so you'd never ever want to use those settings). If you're shooting high-motion high-action high-detail scenes, running through a forest with heavy backlight etc, then definitely the Nano at high bitrates will provide a noticeably sharper image than the onboard codec.
As for the Ninja -- that wouldn't be my first choice for the AF100. For an HDMI-only camera, sure, but I think the benefits of HD-SDI make the Samurai the much better choice, if someone wants to buy an Atomos recorder.
Sometimes I feel like I'm doing the Atomos and Convergent guys a disservice, so let's be clear -- I love love love the Nano. I think it's great. But what I've been saying is that the improvement in image quality, vs. the price, vs. the performance of the native codec on the AF100... I just don't know that it's a necessary expenditure for most AF100 shooters. On an F3, it's a no-brainer -- get it. Duh. Absolutely. The relatively tiny cost, vs. the huge performance increase, is well worth it. But when someone is buying a $4700 body, which already offers a better codec, the amount of difference doesn't necessarily give you that much more bang for your buck.
I think folks should try out the native codec and see if they're happy with it. Most people on this board have come to the same conclusion Hywel has. Some (like KyleProhaska) have decided the opposite. It's an individual choice. It's funny that I'm a Nano user but I'm telling folks that they probably don't need it, but then again, I use $3900 Compact Primes on my AF100 too and I'll be the first to tell you that you can get beautiful footage out of an old $100 used Nikon prime too. So evaluate it based on your own needs. There is no doubt at all that the Nano can deliver better-looking footage. Just decide where your dollars are best spent, is all. If you've already got good lights and good glass and good mics, and are looking for the next step up, by all means get the Nano if you think the onboard codec isn't quite delivering everything you need.
Dan Keaton
Well-known member
Dear Barry,
Nice Post! Nicely worded.
The nanoFlash is not for everyone, nor is it needed in every situation.
It is certainly nice to have a backup, and to be able to record in 4:2:2, and to be able to record at high bit-rates, if desired.
And there are other advantages as the nanoFlash is a very versatile device. But, each and every person should decide if it is an appropriate purchase for them.
Nice Post! Nicely worded.
The nanoFlash is not for everyone, nor is it needed in every situation.
It is certainly nice to have a backup, and to be able to record in 4:2:2, and to be able to record at high bit-rates, if desired.
And there are other advantages as the nanoFlash is a very versatile device. But, each and every person should decide if it is an appropriate purchase for them.
I disagree with this. "Isn't much of a difference" is hard to quantify. To my eye, there is. And clearly the BBC feels the same way. In my early days with the AF before I received the Nano, I found myself very frequently noticing banding, even in first generation footage right off the camera. I see far less now from Nano footage. This is particularly evident in any kind of treatment of the footage to bring it to delivery format, ie DVD or Bluray. The onboard codec does not hold up well to additional compression. Banding tendencies increase dramatically with the introduction of another AVC pass to the edited on board AVC. However, when I compress the I-Frame Nano footage down from the original 140 mb/s footage, there isn't a noticeable increase. The Nano also saved my rear as a stand alone recorder for something I had not planned on needing it to do. It also allows me to send HDMI video out to a consumer client monitor, as well as my 5.6 TV Logic, where otherwise I'd need a much more expensive SDI client monitor.
I've said this numerous times, and it still is hardly mentioned. What exactly is the point in debating how this footage compares first generation? What matters is what it looks like when you deliver it, after you've recompressed it for DVD, Bluray, or file based playback. I see a big difference between Nano and AVC on board when comparing delivered product.
I totally agree that for many users of this camera, there are better ways to spend $3k, and that point is well taken.
Last edited: