With all this talk of 10 Bit for the Action 3, which may or may not happen. It will not make any difference for most people and any positive difference compared to 8 Bit will likely be completely mitigated by certain technical deficiencies, such as the Action 3's sensor and the likely inclusion of 4:2:0 chroma subsampling, plus the issues of using an inter-frame codec compared to a more suitable intra-frame codec.


Here's a video explaining why 10 Bit on the Hero 11 is basically a waste of time. This video deals with the insane claims by many YouTubers that 10 bit on the Hero 11 makes a huge difference compared to 8 Bit. Given that YouTube is an 8 Bit platform for SDR, these claims of “10 Bit is better” for the examples being played back on YouTube are completely wrong. That’s not an opinion, it’s a technical fact.

So, if the Action 3 does get 10 Bit recording ability in the future, for the vast majority of people it will not make any difference what so ever compared to the Action 3’s 8 Bit recordings as far as colour banding etc.

While this example and explanation are dealing with an 8 Bit delivery platform, the exact same outcome will apply with regard local file playback for most people, as they will not be viewing 10 Bit anyway. Also, as said, the issues inherent within such a small “non professional” sensor and limited codec, will also be an issue when previewing in 10 Bit anyway and effectively result in no difference between 10 Bit and 8 Bit acquisitions on such cameras.

Bottom line. If you're happy with your Action 3, don't feel that you are missing out because the Hero 11 has a 10 Bit option.

BTW. Here's something extra that may be useful to anyone intersted in the whole 10 bit vs 8 bit thing.

For anyone interested. I'll be doing another video soon that explains 10 bit vs 8 bit that will help visualise the benefit a lot easier for showing colour banding. Here's a couple of files, one is 10 bit the other is 8 bit. They originate from a 10 project that uses a 10 bit source and 8 bit source of the same colour example, with alternating cuts and split screens that are then exported to 10 and 8 bit.

If you view these clips on a 10 bit device you'll see the difference, if you view on an 8 bit device you won't see any difference. These examples help to show why 10 bit makes no difference what so ever for most people and clearly show colour banding with 10 Bit when played on a 8 Bit device or when converted to 8 Bit, which is what all streaming platforms will do for SDR.

The files are using H.265 for the convenience of small files for downloading. There's also the use of 4:2:0 for these examples, which is fine for "end user delivery". I will also provide 4:2:2 downloads for my  final video and examples, although 10 Bit 4:2:2 H.265 files can be quite difficult to playback for many people, hence the 4:2:0 versions. Some people may also find certain difficulty playing any 10 bit and any H.265. However, 10 Bit has to be used to prove the point and H.265 is the most convenient codec for such examples and also for download bandwidth/speeds.

https://www.mediafire.com/folder/lsoln329nfup2/10_Bit_VS_8_Bit_colour

I think its only reasonable if your color grading.

Well if its being uploaded from the camera sure.... But if theres any color grading/even phone-app based filters etc being used. Certain colors and information won't be lost in those edits. So once they are rendered back and uploaded/compressed whatever. You won't get blown out colors? I'm a noob at post editing softwares but thats my guess?

G1tgudscrub Posted at 10-14 21:13
Well if its being uploaded from the camera sure.... But if theres any color grading/even phone-app based filters etc being used. Certain colors and information won't be lost in those edits. So once they are rendered back and uploaded/compressed whatever. You won't get blown out colors? I'm a noob at post editing softwares but thats my guess?

Rather than taking a “guess” would you like to explain technically what it is you are trying to say?

What do you mean by “blown out” colours?

What phone app are you talking about and what bit-depth does it process at for your “colour grade”?

G1tgudscrub Posted at 10-14 21:13
Well if its being uploaded from the camera sure.... But if theres any color grading/even phone-app based filters etc being used. Certain colors and information won't be lost in those edits. So once they are rendered back and uploaded/compressed whatever. You won't get blown out colors? I'm a noob at post editing softwares but thats my guess?

Rather than taking a “guess” would you like to explain technically what it is you are trying to say?

What do you mean by “blown out” colours?

What phone app are you talking about and what bit-depth does it process at for your “colour grade”?

Hi there. Thank you for reaching out. Osmo Action 3 currently does not support recording and playing 10-bit footage, but supports 8-bit videos. Currently, our engineers are working to make 10-bit color depth supported in Osmo Action 3. Please be patient and stay tuned to DJI's latest news. Thank you for your support and understanding.

@David_Harry While you are certainly correct with your statement as such, it is the same as me saying, most of the people do not need an action cam, smart phone is enough.

I can also follow your argumentation. If you record 16M colors, upload to youtube with 16M colors and view it on a screen that can display 16M colors only, more colors will not be used.

Yet, color banding is a real problem for action shots.https://photographylife.com/what ... g-and-how-to-fix-it
Another case is underexposure, e.g. riding from a sunny area into the deep woods and back. With more colors recorded you can still rescue the scene in post production.

My argument would be, you can always convert an image with 1000M colors to 16M colors (either during post processing or when viewing) but if the source material is limited, you cannot.

And the final argument, even if you do not have HDR today, you will have in future. Will you re-record the same scene then? If you record with 10bit color today, youtube supports HDR videos and your computer/TV will and somebody else's does already. In worst case, you rerender your video tomorrow with HDR output.

Nobody needed 4k cams in 2015 to view it on their FullHD monitor, yet I love watching the same scenes today on the UHD screen.

Yeah, for most it won't be needed.

WernerD Posted at 10-15 08:48
@David_Harry While you are certainly correct with your statement as such, it is the same as me saying, most of the people do not need an action cam, smart phone is enough.

I can also follow your argumentation. If you record 16M colors, upload to youtube with 16M colors and view it on a screen that can display 16M colors only, more colors will not be used.

It would seem that you've not fully understood the points being raised here and don't forget, these are not opinions, it's all based on technical fact.

You said, "it is the same as me saying, most of the people do not need an action cam, smart phone is enough". However, the points I'm making about any low end camera doing 10 Bit vs 8 Bit, especially with the type of chroma subsampling being used, is nothing like your analogy of comparing a phone to an action cam.

As I've clearly said, the problem with these small sensors is that their inherent noise will effectively act like dithering. Therefore negating the difference between 10 bit and 8 bit and especially because of the chroma subsampling.

Add to that the use of an inappropriate codec for the use of 10 bit in post that further diminishes the usefulness of 10 bit on these cameras and further raise the uselessness. There's a very obvious technical reason why you use inter-frame and intra-frame codecs for particular uses, final end delivery and post production.

That banding you see in the sky isn't just the difference between 10 bit and 8 bit, there's also a compounding issue with the codec itself, bit rate, chroma subsampling etc.

I've shot the exact same thing on a Hero 11 in 10 bit and 8 Bit, used a 10 bit pipeline for post, pushed and pulled both for colour, used a 10 bit monitor and there's no obvious difference, due to all the aforementioned issues. However, shooting the same on a better camera/sensor using a better codec, at least an intra-frame 10 Bit 4:2:2 one, such as Pro Res 422 HQ, will yield the types of results you are talking about.

BTW, here's something extra that I've now added to the main post which you may find useful.

For anyone interested. I'll be doing another video soon that explains 10 bit vs 8 bit that will help visualise the benefit a lot easier for showing colour banding. Here's a couple of files, one is 10 bit the other is 8 bit. They originate from a 10 project that uses a 10 bit source and 8 bit source of the same colour example, with alternating cuts and split screens that are then exported to 10 and 8 bit.

If you view these clips on a 10 bit device you'll see the difference, if you view on an 8 bit device you won't see any difference. These examples help to show why 10 bit makes no difference what so ever for most people and clearly show colour banding with 10 Bit when played on a 8 Bit device or when converted to 8 Bit, which is what all streaming platforms will do for SDR.

The files are using H.265 for the convenience of small files for downloading. There's also the use of 4:2:0 for these examples, which is fine for "end user delivery". I will also provide 4:2:2 downloads for my  final video and examples, although 10 Bit 4:2:2 H.265 files can be quite difficult to playback for many people, hence the 4:2:0 versions. Some people may also find certain difficulty playing any 10 bit and any H.265. However, 10 Bit has to be used to prove the point and H.265 is the most convenient codec for such examples and also for download bandwidth/speeds.

https://www.mediafire.com/folder/lsoln329nfup2/10_Bit_VS_8_Bit_colour

Yes, that is the important test. No argumentation beats real life validation.
I've shot the exact same thing on a Hero 11 in 10 bit and 8 Bit, used a 10 bit pipeline for post, pushed and pulled both for colour, used a 10 bit monitor and there's no obvious difference, due to all the aforementioned issues.

DJI does a great job w/o it, so  when they do get it  might as well  have  extra work doing it on an color grading editor

WernerD Posted at 10-16 02:11
Yes, that is the important test. No argumentation beats real life validation.
I've shot the exact same thing on a Hero 11 in 10 bit and 8 Bit, used a 10 bit pipeline for post, pushed and pulled both for colour, used a 10 bit monitor and there's no obvious difference, due to all the aforementioned issues.

Hi Werner.

Here's a link to two Hero 11 files. These are in full frame, 4K/UHD 25FPS, Flat, auto exposure, auto white balance, both high bit rate. One is 10 bit the other is 8 bit.

I'll be doing a YT video about using both in Resolve to show that there's no obvious difference when using the Resolve colour processing. I'll have a download for both the Resolve project and the media but you may be interested in playing with the media with whatever NLE you use.

You'll see what I was talking about with the sensor and codec just not being good enough to record enough clean information to make a meaningful difference in post between the Hero 11's 10 bit and 8 bit recordings.

Given that the Hero 11 has a better sensor than the Action 3 and the Hero 11 isn't good enough, it's easy to see why the Action 3 won't benefit from 10 bit if it does get it. To all intent and purpose, 10 bit on these types of video cameras is at best a placebo and at worst a complete waste of time.

https://www.mediafire.com/folder/wwmc8u87jlf3o/WernerD

I would state that I was able to see a difference in the original video files and that the 10bit file has less bit flips and a more precise line of same colors.
Having said that, the example is not a good one as even the 8bit file does not have immediate apparent color banding but lots of dithering.

The 8bit file


The 10bit file


Thanks for the effort, I really apreciate your help so we can put things in perspective.

WernerD Posted at 10-16 09:54
I would state that I was able to see a difference in the original video files and that the 10bit file has less bit flips and a more precise line of same colors.
Having said that, the example is not a good one as even the 8bit file does not have immediate apparent color banding but lots of dithering.

Hi Werner.

Actually, what you are showing is not the result of 10 Bit or 8 Bit, you are clearly showing the issues I've been talking about with using 10 bit on such a camera. Plus a whole bunch of other nasty stuff inherent within the source codec and sensor etc.

Also. What you have done is not what is considered to be a grade, you've gone way too far with whatever it is you've done. Even a proper 10 Bit 4:2:2 intra-frame source would start showing issues in the chrominance channels with how far you've pushed those colours for typical rec.709 manipulation. So doing what you've done to this type of footage is going to destroy immediately with both the 10 bit and 8 bit sources.

Although, you are proving my point about the sensor and codec issues for such source files. As in, there's far too much noise in the first instance and the codec just isn't suitable.

As for the example not being good because the 8 bit version doesn't have "immediate apparent colour banding", that's neither accurate or any technical way to detirmine chrominance. Both examples suffer the exact same problems, as I've already mentioned. Plus, having noticeable colour banding to start off with isn't how you measure such things. No one takes shots with "deliberate colour banding", it's a byproduct of the bit depth and will be inherent in any 8 Bit footage, regardless of the codec and whether it's objectively noticeable or not from one shot to another.

Given that there was easily enough colour tone variation in the shot, especially the sky, this also proves my point that the camera, sensor and codec are simply not good enough to resolve the relevant and appropriate information to make a 10 bit variation useful in post.

Given that the exposure was perfect for what the camera is capable of, the clips are simply a result of what the camera is capable of in a best scenario situation. Which is something that's fine for the intended purpose of playback and editing but not good enough for proper colour or other destructive post manipulation. The noise off the sensor which gives the effect of dithering, is actually proof of that, in both 10 bit and 8 bit examples.

BTW. What NLE or post software did you use to colour the files? I'm only asking because those stills are a much lower resolution compared to the source files, which suggests that you weren't using the correct project settings for such files. In fact, they look like very similar to a VGA resolution. Plus, the NLE or post application you've used may also have not decoded/renderd the files properly and had other issues with the project settings if you weren't even in the right resolution, as your stills definitely suggest.

As I and others have repeatedly metioned, taking stills from video, especially in a completely different resolution etc. Is not the way to guage or judge video and there are also many different potential issues with taking stills that aren't done correctly.

Here's an example of the same footage being processed and exported correctly. There are two clips being cut between, one is a comp of the original files with a split screen of the 10 bit and 8 file and the other is a coloured variation of the original with the same split of 10 bit and 8 bit.

The colouring here is something reasonable that would be typical of a standard grade. As in, within the bounds of typical rec.709 for a typical output.

For convenience, there's a YouTube version below, which has obviously been converted to 8 bit by YouTube. But here's a link to a 10 Bit render/export of the example, playing this on a 10 bit playback and obviously an 8 bit playback, will show no obvious differences.

https://www.mediafire.com/file/d ... BIT_OUTPUT.mov/file

I pretty sure most modern devices can show 10-bit, or am I missing something?

I don’t think that YouTube sends out 10 bit yet.

if they ever release it enjoy yourgopro, not the same camera as a DJI. and doubt they release it for all the  resolutions either

JamieDC Posted at 11-13 20:50
I pretty sure most modern devices can show 10-bit, or am I missing something?

That's not correct, most modern devices are only 8 bit and all major streaming platforms are only 8 bit for rec.709.

The vast majority of streamed video is done via phones and tablets, these are mostly 8 bit.

Streaming to computers is the same, most computer setups are only 8 bit for their displays, even if the computer itself can handle 10 bit.

Even for people using TVs to monitor back camera files or streaming video, most TVs in the world are 8 bit.

David_Harry Posted at 11-14 05:06
That's not correct, most modern devices are only 8 bit and all major streaming platforms are only 8 bit for rec.709.

The vast majority of streamed video is done via phones and tablets, these are mostly 8 bit.

But it is a big difference for having 8 bit out and working with a higher bitrate file.
It have something to do with dithering.

https://en.wikipedia.org/wiki/Di ... nd_image_processing

johansenfoto Posted at 11-14 05:22
But it is a big difference for having 8 bit out and working with a higher bitrate file.
It have something to do with dithering.

Higher bit rate file has nothing to do with anything.

Dithering is a completely destructive process and adds a noise pattern to any image that it is applied to. Dithering is not a solution, if it were then surely all those Hollywood movies that get encoded to 8 bit would use it, which they don't.

Plus you can already do a similar process anyway to an 8 bit signal.

And, as I've already mentioned, these small sensor cameras have so much sub pixel noise that their sensors effectively add noise similar to dithering, once again demonstrating that 10 bit on such cameras is a complete waste of time.

David_Harry Posted at 11-14 05:38
Higher bit rate file has nothing to do with anything.

Dithering is a completely destructive process and adds a noise pattern to any image that it is applied to. Dithering is not a solution, if it were then surely all those Hollywood movies that get encoded to 8 bit would use it, which they don't.

Point of having 10 bit source or even 12 bit for higher end cameras is for getting most out of the post production.

If you ever worked with images you will have know that working with a 8 bit JPG vs 16 bit tiff is a huge difference. Try recover highlights and shadows on 8 bit.

johansenfoto Posted at 11-14 08:40
Point of having 10 bit source or even 12 bit for higher end cameras is for getting most out of the post production.

If you ever worked with images you will have know that working with a 8 bit JPG vs 16 bit tiff is a huge difference. Try recover highlights and shadows on 8 bit.

The reason for 10bit or any higher bit depth for video is not "for getting most out of post". Depending on the video formats being shot and their intended uses, the bit depth is simply an intrinsict characteristic of the format.

Sure, 10bit+ will give more latitude in post but that's not the reason for shooting in such bit depths. Certain formats simply won't work in 8 bit.

You say "if you ever work with images......." I work with video not JPGs or TIFFs and I'd never use either as an image sequence for a video intermediate. As for recovering highlights and shadows, I'd suggest you concentrate on getting you photos exposed correctly at the lens and then you wouldn't have such issues.

If you have ever worked professionally in TV or cinema post, then you'd understand why 10 bit on an action cam, with all the issues that I've already mentioned, is a complete waste of time.

It's also very obvious that most people within these posts talking about 8 bit and 10 bit do not have a clue about what it is they are talking about. I see a lot of people repeating the same misguided and misunderstood detritus that they've picked up from others who are just as misguided and confused.

You yourself brought dithering in to the conversation without any justification or example for its use and you most certainly didn't give any technical reasoning for it or its drawbacks.

Far too many people on this forum simply do not now enough technically about what it is they are talking about. At least my posts and comments are technically detailed and have examples, that's because I do know what I'm talking about.

If anyone else would like to respond to this post and my comments, then do so with proper technical detain and not a two line, throw away response that just shows your ignorance. And while you are it, show some of your own video examples to prove your point, just like I have.

Bottom line here is this. Giving an action camera 10 bit ability, doesn't make it a Venice, Alexa or Raptor.

Action camera manufacturers should concentrate on making action cameras the best version of what they are and stop trying to make them be something that they aren't.

So on that last point here's a short survey for Action 3 users.

1. Should DJI concentrate on bringing 10 bit (with an inadequate chroma subsampling range for post) to the Action 3?

or

2. Should DJI be concentrating on making the Action 3 work as it should. As in, make sure the thing is in focus (I personally think that having a camera with a lens that's in focus is probably better than having one with a lens that isn't in focus but that's just my opinion )?