AVCHD high bitrate tests- MUD FREE footage

Great stuff, IanB! I was a bit underwhelmed by the Vimeo version but the MTS original looks spot-on. I'm going to patch my camera in a few minutes and go shoot some on my own.

One problem - when I load the MTS file into Cineform NeoScene, it rejects the file. Do you know of a way to transcode from MTS to another format like ProRes? When I've got some test footage I'll try FCP's Log and Transfer as well.

I use clipwrap

http://clipwrap.com/
 
I did some tripod tests.
I wanted to see resolution and color banding in gradients.
Also I included Mjpeg at higher bitrate

--Quality settings (E1 to E4) - 720, 412, 352, 300
--Table settings (E1 to E4) - 24, 24, 24, 24

--For AVCHD 1080 and 720 I used D settings

For Avchd 1080 D settings I got 9734 Kbps in this scene... It isn't sufficient to banding artefacts.It seems that resolution is the same and banding is still present
However I found that Mjpeg has some problems with blacks.. In Ae it has some blue tint in shadows and in player mjpeg too contrasty and seems it has less dynamic range compared to AVCHD 720 or 1080.

Download link:


http://www.nikolicnemanja.com/test6/Normal/1080normal.MTS
http://www.nikolicnemanja.com/test6/Normal/720normal.MTS
http://www.nikolicnemanja.com/test6/Normal/Mjpegnormal.MOV

http://www.nikolicnemanja.com/test6/Dsettings/1080-Dsettings.MTS
http://www.nikolicnemanja.com/test6/Dsettings/720-Dsettings.MTS
http://www.nikolicnemanja.com/test6/Dsettings/Mjpeghi.MOV
 
Last edited:
Same settings as before guys, this time a walk & talk shot with me holding the Steadicam and giving you some skin-tones (and stubble probably!) to pixel peek.

Hey Ian- this is awesome- just downloaded mts. file--looks fantastic I couldn'r find any real classic mud especially when you were making hard pivots.

btw- great 'real-world' run-n-gun example.

It gives you a good variety of texures, i.e. trees, skin tones, lens flares, whites/blacks, motion, pans, hand-held, you name it- it's in there.


It looks like you just shaved....this codec is getting so dang good- it looks like you have to shave again!!! :grin::grin:

Thanks for the efforts
 
I did some tripod tests.
I wanted to see resolution and color banding in gradients.
Also I included Mjpeg at higher bitrate

--Quality settings (E1 to E4) - 720, 412, 352, 300
--Table settings (E1 to E4) - 24, 24, 24, 24

--For AVCHD 1080 and 720 I used D settings

For Avchd 1080 D settings I got 9734 Kbps in this scene... It isn't sufficient to banding artefacts.It seems that resolution is the same and banding is still present
However I found that Mjpeg has some problems with blacks.. In Ae it has some blue tint in shadows and in player mjpeg too contrasty and seems it has less dynamic range compared to AVCHD 720 or 1080.

Download link:


http://www.nikolicnemanja.com/test6/Normal/1080normal.MTS
http://www.nikolicnemanja.com/test6/Normal/720normal.MTS
http://www.nikolicnemanja.com/test6/Normal/Mjpegnormal.MOV

http://www.nikolicnemanja.com/test6/Dsettings/1080-Dsettings.MTS
http://www.nikolicnemanja.com/test6/Dsettings/720-Dsettings.MTS
http://www.nikolicnemanja.com/test6/Dsettings/Mjpeghi.MOV


Didn't Tester13 mention that there was something strange with the D settings or am I mistaken?
 
These are settings that tester used:
Video buffer - a) 33.500.000 b) 40.000.000
Video bitrate adjustment - a) 22.000.000 b) 32.000.000
Overall bitrate adjustment - a) 24.000.000 b) 35.000.000
Limiting bitrate adjustment - a) 26.000.000 b) 40.000.000


Do any of these have an effect on mjpeg mode? How do we adjust buffer etc for mjpeg?


.
 
As for test results using low level analysys:

Only D setting is ok at low level.
All others contains large waste of space as your bitrate is consumed in garbage large size spike frame.
As Tester said...
 
These seem good for now!

These seem good for now!

I tried with this settings and it's working.
I record 4 minutes long clip ( I stopped the camera)
Video buffer 40.000.000
Video bitrate adjustment 40000000
Overall bitrate adjustment 46000000
Limiting bitrate adjustment 50000000

Overall bit rate : 34.8 Mbps
Maximum Overall bit rate : 46.0 Mbps

Other settings, incl. d(suggested as possibly a good set by tetser13) are as follows;
Video buffer 40.000.000
Video bitrate adjustment - c)40000000 d)50000000 e)60000000 F)55000000
Overall bitrate adjustment - c)46000000 d)56000000 e)66000000 F)57000000
Limiting bitrate adjustment - c)50000000 d)60000000 e)70000000 F)63000000
 
I tried with this settings and it's working.
I record 4 minutes long clip ( I stopped the camera)
Video buffer 40.000.000
Video bitrate adjustment 40000000
Overall bitrate adjustment 46000000
Limiting bitrate adjustment 50000000

Overall bit rate : 34.8 Mbps
Maximum Overall bit rate : 46.0 Mbps

Other settings, incl. d(suggested as possibly a good set by tetser13) are as follows;
Video buffer 40.000.000
Video bitrate adjustment - c)40000000 d)50000000 e)60000000 F)55000000
Overall bitrate adjustment - c)46000000 d)56000000 e)66000000 F)57000000
Limiting bitrate adjustment - c)50000000 d)60000000 e)70000000 F)63000000

In order to see if settings are good you need to analyse with elecard software. Then you should see bitrate at low level.
As for settings c , e and f they all have spikes and they are not good( they are good but your wasting your card capacity).
Only d is good at low level...
 
Before I went to bed last night I checked the originally posted before-and-after example and try as I might, at normal speed or when stepping through frame by frame, I couldn't see any persuasive difference between the two - certainly I wouldn't be able to pick which is which if presented with the two without prior knowledge of their identity. I've just done the same in the clear light of day, and I still can't. Maybe it's not a suitable example, or there's something wrong with my eyes.

Has anything else been posted before-and-after which would be more persuasive? Just 'after' isn't a meaningful test in my book.
 
One problem - when I load the MTS file into Cineform NeoScene, it rejects the file. Do you know of a way to transcode from MTS to another format like ProRes? When I've got some test footage I'll try FCP's Log and Transfer as well.

Rename your mts files from lower case mts to uppercase and that should fix the neoscene issue.
 
I updated my camera with the setting B firmware for a PAL camera (also did the panasonic battery patch and 30min limit patch) and the time remaining for FHD is still 1 hour on an 8 gig card, but MJPEG HD has dropped right done to 8 mins (from the 8gig 33 mins stated in the manual)

New AVCHD footage looks to be MUD free, and at the higher data rate.

What's going on?
 
I updated my camera with the setting B firmware for a PAL camera (also did the panasonic battery patch and 30min limit patch) and the time remaining for FHD is still 1 hour on an 8 gig card, but MJPEG HD has dropped right done to 8 mins (from the 8gig 33 mins stated in the manual)

New AVCHD footage looks to be MUD free, and at the higher data rate.

What's going on?



8 GB.....
8 minutes of true HD .....

sounds about right to me
 
But it's the AVCHD FHD, time remaining, which should change? not the MJPEG HD for this firmware change?
 
Has anything else been posted before-and-after which would be more persuasive? Just 'after' isn't a meaningful test in my book.
Well, jobless posted some footage in the first post. Looking at frame extracts the difference is quite clear (to my eyes). http://www.dvxuser.com/V6/showpost.php?p=2012714&postcount=12
Unpatched output is full of smeared compression artifacts, while patches has smooth "proper" motion blur. The differences are only evident when motion is present in the scene. Tripod tests (http://www.dvxuser.com/V6/showpost.php?p=2013231&postcount=63) produce identical footage (too my eyes).
 
why are you using cineform? just to have an intermediate codec? the pulldown is already done in camera.

Yeah, just for transcoding. Easier than launching FCP if I just want to convert a few files (and FCP requires you keeping all the other directories/files from the card, rather than just the MTS).

Shot some footage with Tester's high quality 1080p AVCHD settings and the native 24p patch. Those files worked fine in Cineform Neoscene, so I don't know what the issue was with the other MTS files. EDIT: Just saw Isaac's post, thanks.

Results look great for the most part, I'll upload a few MTS's for you guys to check out in a few minutes.
 
Back
Top