Inter-frame- vs. Intra-frame compression

[xQubeZx]

2 minutes ago, .spider. said:

No

 

Nobody is stopping you from doing your own analysis. 

As I in reality don't give a shit about the subject I wont. However you can't say this is a scientific study worth trusting in. 

[AkiraDaarkst]

23 minutes ago, xQubeZx said:

Okey, I'm not a pro at any of this stuff but shouldn't RAW footage be used for comparisons like this? An already compressed video will have discared important information and probably made the footage better suit one type of compression? And isn't 5mbit/s quite low for something like this?

 

Isn't this like taking a 60fps clip, exporting it in 24fps and then trying to scale it up to 60fps again and expecting you to get the full 60fps you had from the start? 

 

Sure I don't know that much on video compression and so on but I do know that you have to be very careful in every scientifc study you do as I have done lots of physics and chemistry labs. Firstly you should start with a theory, then back it up with mathematical or similar proofs and then finally do several experiments to prove it. You should state how you do it, what you use (this is very important so other can do the tests themselves to verify the "discovery". You should have exact numbers and hardware and material specs). No offence but to me this seems poorly made and has a nothing to really back it up as it lacks a lot of info and proper techniques to actually prove something. (Several video sources and different recording formats for example?)

In the acquisition stage, as in recording with the camera, Intraframe will generally be better because cameras that support both Intraframe and Interframe compression gives more bit rate headroom to the Intraframe version.

 

In the delivery stage, when you render out the project to be posted on YouTube or made into a DVD, it's better to go for Interframe compression because you want the files to be as small as possible while retaining quality.  The final delivery product has gone through stuff like color grading, noise reduction, warp stabilization, and other edits and cuts.  Interframe masters do not really handle such editing as well as Intraframe masters.  This is why I also said previously, only a fool would render an Intraframe version that has the same bit rate as an Interframe version.

 

He's only comparing the delivery side of things.

 

Sony for example, in their famous FS7, XAVC-I (which is Intraframe) records at a whopping 600Mb/s at 4K 60p and only 150Mb/s in XAVC-L (Interframe), so NO, Interframe is not always better than Intraframe.  Same with Canon, on their C300 MK II, Intraframe compression records up to 410Mb/s and Interframe is only 50Mb/s.

 

The Canon 5D camera with All-I and IPB on the other hand, the IPB (Interframe) can be better than All-I (Intraframe) because the bit rate headroom of All-I is only slightly better than IPB.

[.spider.]

11 minutes ago, xQubeZx said:

As I in reality don't give a shit about the subject I wont. However you can't say this is a scientific study worth trusting in. 

I have never said that it is a scientific study, 

but it shows pretty well that this:

 

On 14.12.2016 at 7:36 AM, AkiraDaarkst said:

They were taken from 720p resolution proxy files which were created using a GOP Inter-Frame compression, if I had used Intra-Frame compression the quality would be better of course.

is a wrong generalization. Since Intra-Frame compression doesn't not improve quality. 

[AkiraDaarkst]

12 minutes ago, xQubeZx said:

As I in reality don't give a shit about the subject I wont. However you can't say this is a scientific study worth trusting in. 

It's not a scientific study if the test methods are already biased towards the favorable outcome.

 

5 minutes ago, .spider. said:

is a wrong generalization. Since Intra-Frame compression doesn't not improve quality. 

Did it ever occur in that tiny little brain of yours that if I wanted to create an Intraframe proxy of the footage I filmed with the RED camera I would use ProRes which would have much higher bit rates?  I've been working in cinematography for years, I know what I am doing.

[.spider.]

5 minutes ago, AkiraDaarkst said:

 

Sony for example, in their famous FS7, XAVC-I (which is Intraframe) records at a whopping 600Mb/s at 4K 60p and only 150Mb/s in XAVC-L (Interframe), so NO, Interframe is not always better than Intraframe.  Same with Canon, on their C300 MK II, Intraframe compression records up to 410Mb/s and Interframe is only 50Mb/s.

 

The Canon 5D camera with All-I and IPB on the other hand, the IPB (Interframe) can be better than All-I (Intraframe) because the bit rate headroom of All-I is only slightly better than IPB.

You are mixing bitrates and encoding schemas, you can only compare encoding schemas at given bitrates. 

[AkiraDaarkst]

21 hours ago, AkiraDaarkst said:

Intraframe does not have a quality advantage if the video has the same bit rate as the Interframe version, it may even look worse.  However only a fool would use Intraframe and set the encoding to the same (as Interframe) or too low a bit rate.

@xQubeZx

I'm quoting something I said, which he ignored because it was inconvenient in his aim to prove that I am wrong.

[xQubeZx]

8 minutes ago, AkiraDaarkst said:

It's not a scientific study if the test methods are already biased towards the favorable outcome.

Yes, but even if it didnt have a favorable outcome it wouldn't be called a study nor something trustworthy. 

 

10 minutes ago, .spider. said:

I have never said that it is a scientific study, 

but it shows pretty well that this:

 

is a wrong generalization. Since Intra-Frame compression doesn't not improve quality. 

Sure you did not say it was a scientifc study, but you can't just go around and tell people what is right and what is wrong in a very technical matter as this when you have nothing to back you up. Then it's just rumors or a self made "theory". 

 

Everything scientifc or similar I read I assume to be false untill someone proofs it otherwise with some real facts and sources.  

Edited December 15, 2016 by xQubeZx

[.spider.]

6 minutes ago, AkiraDaarkst said:

It's not a scientific study if the test methods are already biased towards the favorable outcome.

My test method wasn't biased to anything. 

 

 

Quote

Did it ever occur in that tiny little brain of yours that if I wanted to create an Intraframe proxy of the footage I filmed with the RED camera I would use ProRes which would have much higher bit rates?  I've been working in cinematography for years, I know what I am doing.

Now the insults are starting @ALwin behaved similar. 

And the quality would be better because of the higher bitrate and not because of the intra-frame compression. Someone who is working in cinematography for years should know that. 

[.spider.]

8 minutes ago, xQubeZx said:

 

Sure you did not say it was a scientifc study, but you can't just go around and tell people what is right and what is wrong in a very technical matter as this when you have nothing to back you up. Then it's just rumors or a self made "theory". 

I have posted values of my analysis what did the other person post so far? 

[xQubeZx]

3 minutes ago, .spider. said:

I have posted values of my analysis what did the other person post so far? 

The point is not what the other person has to say as you don't have anything either. You have some vauge values. The original post doesn't even include what clip was used. Notihing about what equipment (Camera) or settings was stated (compression, shutter, ISO, lights and so on. It may not matter but should be stated eitherway). There is no theory stated on what you expect to happen and there is no final discussion discussing your values and how and why they turned out that way. 

 

However what annoys me most is the use of only one test. The same test should be done several times and several similar tests with different settings should be used. There you will have some real data you can plot in excel and create some proper diagrams actually explaining what happens and how everything is related. 

 

So even if he hasn't posted any proof (yet?), you haven't either. 

 

But basically I wouldn't reccomend anyone to trust this before any real study or proof has been made as so far its a mere statement without a underlaying base. Try to say to yourself that this is more than enough info to accuratly belive in something but any sane person wouldn't belive so heavily in this. (Personally I would belive there is a time and use for both "formats" with their separate advantages)

[xQubeZx]

Here you have a good PDF from MIT I think you should read before posting an analasys. 

 

https://ocw.mit.edu/courses/biological-engineering/20-109-laboratory-fundamentals-in-biological-engineering-fall-2007/assignments/sci_writing_guid.pdf

[.spider.]

8 minutes ago, xQubeZx said:

Here you have a good PDF from MIT I think you should read before posting an analasys. 

 

https://ocw.mit.edu/courses/biological-engineering/20-109-laboratory-fundamentals-in-biological-engineering-fall-2007/assignments/sci_writing_guid.pdf

I am not doing my PhD here. 

[xQubeZx]

Just now, .spider. said:

I am not doing my PhD here. 

You are not posting anything trustworhty either. 

[.spider.]

15 minutes ago, xQubeZx said:

 

However what annoys me most is the use of only one test. The same test should be done several times and several similar tests with different settings should be used.

Actually 1400 frames at 6 different settings were tested.

[xQubeZx]

Just now, .spider. said:

Actually 1400 frames at 6 different settings were done.

It doesn't say that anywhere. But as I've got it from the very small amount of information that would be 6 different compression settings as you have still only told one "raw" video source. Not like 10 different ones in different conditions with different cameras and compressions from the start.

[.spider.]

1 minute ago, xQubeZx said:

It doesn't say that anywhere. 

If you look at the graph you can clearly see 5 data points per metric, The 6th is not plotted because that was the source compared with the source.

[xQubeZx]

1 minute ago, .spider. said:

If you look at the graph you can clearly see 5 data points per metric, The 6th is not plotted because that was the source compared with the source.

Sure, okey I missed that, but really there is a serious lack of info that you can't cover up now. A bit to late for that.

[AkiraDaarkst]

17 minutes ago, xQubeZx said:

Personally I would belive there is a time and use for both "formats" with their separate advantages

There are uses for both.

1. Generally, cameras that support both Intraframe and Interframe codecs/compression, the Intraframe version records in higher bit rates.
2. Even codecs which have both Intraframe and Interframe versions, the Intraframe version supports higher bit rates.

However, even if a video was encoded to very similar or same bit rates into Intraframe and Interframe versions, depending on what is in the scene and the motion that exists or doesn't exist, one will be better than the other or both will be visually the same or very similar until the editing begins and one will fall apart faster than another.  What I like about Intraframe is that each frame is independently compressed instead of being grouped together with neighboring frames and then compressed.  When you start cutting footage, Intraframe versions are generally preferred because you can cut at any frame, while with Interframe when you start cutting you may be cutting on a P or B frame which requires retrieving data from an I-frame.

 

What Interframe does is, it takes 1 key frame and when it looks at the next frame it only compresses the changes that happened from the previous frame.

[xQubeZx]

2 minutes ago, AkiraDaarkst said:

There are uses for both.

1. Generally, cameras that support both Intraframe and Interframe codecs/compression, the Intraframe version records in higher bit rates.
2. Even codecs which have both Intraframe and Interframe versions, the Intraframe version supports higher bit rates.

However, even if a video was encoded to very similar or same bit rates into Intraframe and Interframe versions, depending on what is in the scene and the motion that exists or doesn't exist, one will be better than the other or both will be visually the same or very similar until the editing begins and one will fall apart faster than another.  What I like about Intraframe is that each frame is independently compressed instead of being grouped together with neighboring frames and then compressed.  When you start cutting footage, Intraframe versions are generally preferred because you can cut at any frame, while with Interframe when you start cutting you may be cutting on a P or B frame which requires retrieving data from an I-frame.

 

What Interframe does is, it takes 1 key frame and when it looks at the next frame it only compresses the changes that happened from the previous frame.

So interframe can be kind of put similar to motionblending in Ae? Or well to a certain extent. And Intraframe can be compared to say a jpeg sequense with an completley induvidal image for each frame.

[AkiraDaarkst]

4 minutes ago, xQubeZx said:

So interframe can be kind of put similar to motionblending in Ae? Or well to a certain extent. And Intraframe can be compared to say a jpeg sequense with an completley induvidal image for each frame.

A close analogy, though not quite the exact same thing.

 

Also, the quality of each type depends on the quality and efficiency of the codec implementation.  I've known cameras with very high bit rate Intraframe codecs that produced worse image than the Interframe footage they captured because the manufacturer screwed up... but a firmware can fix the issues.  And I've also know cameras with high bit rate Interframe codecs that were just seriously bad that similar bit rate Intraframe version videos were better.

[AkiraDaarkst]

19 minutes ago, xQubeZx said:

 

Anytime you see a codec being described as GOP (Group of Pictures), you know it's going to likely use Interframe compression.  So for Sony XAVC-L means Long GOP and XAVC-S means Short GOP   my mistake here, XAVC-S is also Long GOP.

https://en.wikipedia.org/wiki/Group_of_pictures

[.spider.]

12 minutes ago, AkiraDaarkst said:

What I like about Intraframe is that each frame is independently compressed instead of being grouped together with neighboring frames and then compressed.  When you start cutting footage, Intraframe versions are generally preferred because you can cut at any frame, while with Interframe when you start cutting you may be cutting on a P or B frame which requires retrieving data from an I-frame.

You like something that doesn't matter anymore because PCs and Disks are fast enough?

 

15 minutes ago, AkiraDaarkst said:

What Interframe does is, it takes 1 key frame and when it looks at the next frame it only compresses the changes that happened from the previous frame.

From the previous and future frames. 

[Omon_Ra]

9 minutes ago, xQubeZx said:

So interframe can be kind of put similar to motionblending in Ae? Or well to a certain extent. And Intraframe can be compared to say a jpeg sequense with an completley induvidal image for each frame.

Kind of. Think of intraframe as how a film camera captures images; each individual frame is being exposed. My GH2 is hacked with an intraframe patch, the creator of the patch puts out an all i-frame/intraframe and a (typically) GOP3 for interframe folks. The bitrates are well above stock bitrates for either and the difference between the intra and interframe footage is barely noticeable; it's really dependent on the user's preferences and SD card speed. If people so desire, I can do a little test shoot between two identical patches, one intra, one inter and yall can decide.

[AkiraDaarkst]

14 minutes ago, Omon_Ra said:

Kind of. Think of intraframe as how a film camera captures images; each individual frame is being exposed. My GH2 is hacked with an intraframe patch, the creator of the patch puts out an all i-frame/intraframe and a (typically) GOP3 for interframe folks. The bitrates are well above stock bitrates for either and the difference between the intra and interframe footage is barely noticeable; it's really dependent on the user's preferences and SD card speed. If people so desire, I can do a little test shoot between two identical patches, one intra, one inter and yall can decide.

This depends on how the Intra and Inter encoding were implemented, if both are well implemented then visually the differences are hardly noticeable.  However there is also the condition of how one wants to edit.  Sony and Canon (their cinema cameras with XAVC or XF AVC Intra and Inter), the Intra versions will always be better than the Inter versions when you want to do heavy editing and color grading.  On cameras like the Canon 5D's with All-I and IPB, the All-I Intra isn't much better than the IPB Inter because Canon limited the bit rate.

 

I've done indoor studio controlled test shots with both a Canon C300mk II and Sony FS7, and the Intra versions hold up much better during editing.  The Inter versions start to show macroblocking during grading.

[Omon_Ra]

5 minutes ago, AkiraDaarkst said:

-snip-

By 'users preferences' I was referring to their editing preferences; I suppose I should have made that more clear.