PDA

View Full Version : GF1 100mb/s footage acheived!!!



mike797
06-26-2010, 02:52 AM
108mb/s video on the GF13 http://vimeo.com/12854862

99.2mb/s video on the GF13 (2min 16s) http://www.youtube.com/watch?v=lk2nj5vBPac

zcream
06-26-2010, 07:55 AM
what card mike ?

mike797
06-26-2010, 08:35 AM
what card mike ?
Class 4 sandisk 16gb ultra ll

blues
06-26-2010, 09:25 AM
what are the settings and wat ptools version

mike797
06-26-2010, 09:29 AM
what are the settings and wat ptools version
Settings are as follows. ptools 3.7d

E1 Q 545, 500, 455, 410
E1 T 24, 24, 24, 24

MJPEG 1280m height checked (720)
MJPEG 1280m width checked (1280)

420->422 checked

sammysammy
06-26-2010, 10:29 AM
wow mike your getting over 100mb with out the camera stop recording?how long can u record at that rate? i have what seems to be the best card the extreme class 10 16bg, and as soon as i reach 80's with 422 checked and the gh1 stops recording because of card speed..sometimes if i pan fast with 422 at about 75mb the camera stops..i have had over 100mb recorded but with our 422 and only for about 5 sec before it stops..all this at 720p..

Exact
06-26-2010, 10:45 AM
How are you measuring the bit rate? What tool are you using ?

Brian@202020
06-26-2010, 10:49 AM
wow mike your getting over 100mb with out the camera stop recording?how long can u record at that rate? i have what seems to be the best card the extreme class 10 16bg, and as soon as i reach 80's with 422 checked and the gh1 stops recording because of card speed..sometimes if i pan fast with 422 at about 75mb the camera stops..i have had over 100mb recorded but with our 422 and only for about 5 sec before it stops..all this at 720p..

Mike has a GF1. Altho the GF1 is less expensive than the GH1, it could potentially have better hardware because it's newer. So the bottleneck might not be the camera like it is with the GH1.

mike797
06-26-2010, 11:08 AM
How are you measuring the bit rate? What tool are you using ?

Using QuickTime and the data rate says 108mb/s

mike797
06-26-2010, 01:43 PM
wow mike your getting over 100mb with out the camera stop recording?how long can u record at that rate? i have what seems to be the best card the extreme class 10 16bg, and as soon as i reach 80's with 422 checked and the gh1 stops recording because of card speed..sometimes if i pan fast with 422 at about 75mb the camera stops..i have had over 100mb recorded but with our 422 and only for about 5 sec before it stops..all this at 720p..
Can record for as long as it allows me to record so 3mins 46sec.
I have noticed when i first turn the camera on and hit record it tries to record complex scenes at 116-118mb/s but cuts out after 2-3secs. I then hit record again and it seems to switch to a lower mb/s of around 102-108mb/s and remains stable until I turn the camera off then it does the above again.

If one plays with these settings below (which i have not ticked in my settings) could that prevent it from attempting to record over 108mb/s?

VBitrateAd. H:
VBitrateAd. L:
overall Bitrate:
Limiting Bitrate:

Kholi
06-26-2010, 01:58 PM
THat's insane. Footage looks very clean and clear.

Nice job testing, looking forward to real 24p Patch for the MJPEG

mike797
06-26-2010, 02:05 PM
THat's insane. Footage looks very clean and clear.

Nice job testing, looking forward to real 24p Patch for the MJPEG

How long do you think it will take for it to be released? Would 24p cause the mb/s to go up or down?

sparedog
06-26-2010, 04:29 PM
If one plays with these settings below (which i have not ticked in my settings) could that prevent it from attempting to record over 108mb/s?

VBitrateAd. H:
VBitrateAd. L:
overall Bitrate:
Limiting Bitrate:


mike, these settings are for avchd i believe and dont have any effect on mjpeg.

check the gf1 testing thread for info on table settings.



I have noticed when i first turn the camera on and hit record it tries to record complex scenes at 116-118mb/s but cuts out after 2-3secs. I then hit record again and it seems to switch to a lower mb/s of around 102-108mb/s and remains stable until I turn the camera off then it does the above again.


this is really interesting, and certainly needs more testing.

sparedog
06-26-2010, 05:25 PM
Can record for as long as it allows me to record so 3mins 46sec.
I have noticed when i first turn the camera on and hit record it tries to record complex scenes at 116-118mb/s but cuts out after 2-3secs. I then hit record again and it seems to switch to a lower mb/s of around 102-108mb/s and remains stable until I turn the camera off then it does the above again.


so i took your lowest setting, q410, t 24 and set
all quality set to 410
all table set to 24

this appears stable on a class 10, giving me clips from 57.20 to 97.80, average about 90mbits/sec!

i need to wait till daylight to test thoroughly. i shall test this one thorougly, as a 100 percent stable codec is what i'm after,
not one that fails to write at that critical moment:badputer:

if anyone has a panny pancake 20mm lens, pleae test this setting, as this lens has a better resolution than mine so i think would force the codec with more detail. Cheers, Huw

alignment1
06-26-2010, 07:00 PM
Using QuickTime and the data rate says 108mb/s

wait, I'm confused...QT can't read MTS files so are these your transcoded Prores files that your getting your readings from?? If so than the data rate is false- am I wrong about this??

Brian@202020
06-26-2010, 08:12 PM
wait, I'm confused...QT can't read MTS files so are these your transcoded Prores files that your getting your readings from?? If so than the data rate is false- am I wrong about this??

MJPEG files are .MOV, the AVCHD files are .MTS. He is achieving that data rate in MJPEG mode and reading it in Quicktime. That's how I've been reading my MJPEG files as well. I use either Toast Video Player or VLC to view the data rates of the AVCHD files.

sparedog
06-26-2010, 11:30 PM
yes quicktime on os x shows you this data for .mov, and it is an accurate straight from camera bitrate.

the 410 24 setting where all q is set to 410 and all t is set to 24, with 4:2:2 selected isn't 100 percent stable, but boy is it nice! in daylight 100mbits on nearly everything!!!

the weird thing is that sometimes it trips the write speed error with a lower bitrate (100.40mbits) and then writes fine for slightly higher bitrates (100.68mbits) immediately afterwards.

PappasArts
06-27-2010, 12:48 AM
yes quicktime on os x shows you this data for .mov, and it is an accurate straight from camera bitrate.

the 410 24 setting where all q is set to 410 and all t is set to 24, with 4:2:2 selected isn't 100 percent stable, but boy is it nice! in daylight 100mbits on nearly everything!!!

the weird thing is that sometimes it trips the write speed error with a lower bitrate (100.40mbits) and then writes fine for slightly higher bitrates (100.68mbits) immediately afterwards.

Run those settings with 4:2:2 unchecked ( regular Mjpeg at hibit rates already has good color ) and then shoot the star chart, see what happens- If it doesn't cause the write error, then these settings are very good. 100% error free is the goal, and when we get the numbers right- that will be awesome. The more people trying- do experiments, the faster we will find the stable formula. However we need to be filming the same testing parameters- that's why the star chart keeps the testing at a standard.

Pappas

mike797
06-27-2010, 02:07 AM
I have been using the 1.7 panny lens

mike797
06-27-2010, 02:56 AM
Ok i just changed all Q settings to 545 and everything failed!

It gave me 2 sec clips at 120mb/s when i filmed my computer screen.

Going to try lower them.

mpgxsvcd
06-27-2010, 03:21 AM
Ok a few more clips using mike's settings. You can download the original files of these as well. The streaming version just doesn't do them justice. Keep in mind that the second file is 1.6 GB though!:Drogar-Shock(DBG):

http://vimeo.com/12878669

http://vimeo.com/12880009

PappasArts
06-27-2010, 03:43 AM
Ok a few more clips using mike's settings. You can download the original files of these as well. The streaming version just doesn't do them justice. Keep in mind that the second file is 1.6 GB though!:Drogar-Shock(DBG):

http://vimeo.com/12878669

http://vimeo.com/12880009


Uncheck 4:2:2- you'll get a better chance of less write errors- Also at this high of a bit rate
the color is already great with Mjpeg. Mjpeg is just jpeg photos in a motion wrapper in plain terms.
When was the last time we thought whether our DSLR taken photos in Jpeg mode were 4:2:2 or not. Anyhow,
give it a try, you might get better card writing results.

Looking forward to seeing if these settings will even work with GH1.

Pappas

rawfa
06-27-2010, 05:30 AM
If it's not asking too much it would be nice to see some skin tones in these testes. Also, close ups with a lot of dof are harder to evaluate than wider framed scenes with more information and diferent light conditions (this way we can evaluate dynamic range, detail, color reproduction, etc)

rawfa
06-27-2010, 05:36 AM
Ok a few more clips using mike's settings. You can download the original files of these as well. The streaming version just doesn't do them justice. Keep in mind that the second file is 1.6 GB though!:Drogar-Shock(DBG):

http://vimeo.com/12878669

http://vimeo.com/12880009

On the last scene on the last video from left to right you can see a house, the sky, a tree and bushes. This is a good frame to evaluate dynamic range. You can see that the highlights on the sky and the white house are completely burned. Now, since this was shot at 108mb/s it should give you a lot of head room for recovering those highlights. Could you try to do that to see how much CC can the footage take before falling apart?

sparedog
06-27-2010, 05:44 AM
and then shoot the star chart,
Pappas

papas, do you have a link to the star chart, when i search for it, i get a whole bunch of dead leads.

i cant say for the others, but i'm really keen to get the 422 to work as i want to use the gf1 for broadcast, but i will shoot the chart with both 420 & 422.

thanks

sparedog
06-27-2010, 09:51 AM
If it's not asking too much it would be nice to see some skin tones in these testes. Also, close ups with a lot of dof are harder to evaluate than wider framed scenes with more information and diferent light conditions (this way we can evaluate dynamic range, detail, color reproduction, etc)

rawfa,

hopefully this footage will fullfill some of your request. I would like to have had a more varied sky for you to play with, but unfortunately we dont have any clouds here in barcelona :grin:

download the original files from vimeo

109mbits obtained with the 410 24 settings. 14-45 panasonic - a lens i don't try to use because it distorts everything for me :(

http://www.vimeo.com/12891887

12891887

onion
06-27-2010, 11:01 AM
If it's not asking too much it would be nice to see some skin tones in these testes.

Whatever floats your boat :grin:

PappasArts
06-27-2010, 01:08 PM
papas, do you have a link to the star chart, when i search for it, i get a whole bunch of dead leads.

i cant say for the others, but i'm really keen to get the 422 to work as i want to use the gf1 for broadcast, but i will shoot the chart with both 420 & 422.

thanks


Sure, here you go.

Tip- when focusing on the star chart get the lines to a crispy jaggie dance. Also move the camera too in separate tests- to simulate change of scene doing circular motions as well swaying it, this will get the VBR going.
Shoot a Fstop at least f8- Gain the camera up and set the Detail to +2- This will force the codec to spike at it's peak more. This process is good for fine tuning the exact formula numbers towards the threshold before write error.

Here is a link to a file the 2k file- use this one
http://www.sendspace.com/file/scn6oz


I also uploaded a picture so you know what I'm talking about. Don't use this one, as it's just a sample- Use the one I uploaded

Pappas

sparedog
06-27-2010, 02:06 PM
papas,

filmed the chart many times trying to trip the error message. when it did finally, i just carried on and filmed some more. the error message has very little effect i can see on the 410 24 settings as the error was tripped at 106.06mbits and the next clipped filmed was 105.33 which continued till i stopped it at 30 second later

some mbits from the charts
105.65 - recorded successfully for about 30 seconds till i stopped it
106.01 - recorded successfully for about 30 seconds till i stopped it
106.06 - tripped error message, the only one i got in 20 clips of 30 seconds
105.33 - recorded successfully for about 30 seconds till i stopped it
104.06 - recorded successfully for about 30 seconds till i stopped it
103.96 - recorded successfully for about 30 seconds till i stopped it
103.36 - recorded successfully for about 30 seconds till i stopped it
104.45 - recorded successfully for about 30 seconds till i stopped it
101.92 - recorded successfully for about 30 seconds till i stopped it

gonna try q 400 with t22, as i really want to get the minimum bitrate up to above 80mbits whilst keeping max at 100mbits. i've hacked my camera 16 times just to do the table tests, i hope there's no limit lol

cheers, huw

sparedog
06-27-2010, 02:36 PM
no 400 20 is not stable :crybaby:

398 22 is not stable

400 24 is not stable either

it seems as with mike797's setting and with 410 24, these become stable once the first error message has become tripped.

therefore i am setting my camera to 410 24 for now. 100mbit 422 is the same specs as DVCPRO HD which is used as the main camera format on many discovery, nat geo channel, history shows :D

in fact, does anyone know of any reason (moire, resolution, etc) the GF1 100mbit 422 footage would not pass broadcast standards as main camera footage, if for example i wanted to make an underwater documentary using gf1s in housings?

HHL
06-27-2010, 04:29 PM
in fact, does anyone know of any reason (moire, resolution, etc) the GF1 100mbit 422 footage would not pass broadcast standards as main camera footage, if for example i wanted to make an underwater documentary using gf1s in housings?


Thanks so much for posting the vid. I'm watching this one VERY CLOSELY....I have a GF1 (as well as a GH1). I'm excited to see this....

Quick request: Could you please post a screen capture? It is really hard to see anything on the Vimeo video. I'm not saying that it's not impressive...I am saying that it's hard to tell if it's in focus....and/or what is Vimeo compression issues, etc...

Thanks again for posting!

sparedog
06-27-2010, 05:11 PM
sure here a grab - i'm not sure if the focus is spot on in the clip lol

and below it is the highest bitrate i've recorded successfully, but the codec would fail too often to be useable.

http://www.huwjenkins.com/109mbits.jpg
http://www.huwjenkins.com/149MBIT.jpg

PappasArts
06-27-2010, 06:28 PM
papas,

filmed the chart many times trying to trip the error message. when it did finally, i just carried on and filmed some more. the error message has very little effect i can see on the 410 24 settings as the error was tripped at 106.06mbits and the next clipped filmed was 105.33 which continued till i stopped it at 30 second later

some mbits from the charts
105.65 - recorded successfully for about 30 seconds till i stopped it
106.01 - recorded successfully for about 30 seconds till i stopped it
106.06 - tripped error message, the only one i got in 20 clips of 30 seconds
105.33 - recorded successfully for about 30 seconds till i stopped it
104.06 - recorded successfully for about 30 seconds till i stopped it
103.96 - recorded successfully for about 30 seconds till i stopped it
103.36 - recorded successfully for about 30 seconds till i stopped it
104.45 - recorded successfully for about 30 seconds till i stopped it
101.92 - recorded successfully for about 30 seconds till i stopped it

gonna try q 400 with t22, as i really want to get the minimum bitrate up to above 80mbits whilst keeping max at 100mbits. i've hacked my camera 16 times just to do the table tests, i hope there's no limit lol

cheers, huw

Excellent testing- The Star chart is really good for this.

The right numbers are important- it appears that the right combination is the key. The more people doing it, the more we learn...

Has anyone done these settings with the GH1 yet- filming the start chart?

Pappas

PappasArts
06-27-2010, 06:30 PM
it seems as with mike797's setting and with 410 24, these become stable once the first error message has become tripped.

I have noticed this even with my settings. A first error- and after that no more. Maybe a buffer needs to purge out or?

Pappas

sparedog
06-28-2010, 01:16 AM
Ok i just changed all Q settings to 545 and everything failed!

It gave me 2 sec clips at 120mb/s when i filmed my computer screen.

Going to try lower them.

mike,

my calculations say for 545, you are writing about 140mbits/sec so you need a table setting of above 32, but then again, i was never very good at maths at school lol

today, i will try all t settings set to 35 with 545 and let you know

huw

sparedog
06-28-2010, 01:46 AM
ok tested 540 with 40.

the write error was tripped alot, even without switching off the camera. however, on the simpler test subjects and when the star chart was not full screen, i got up to a reading of 108mbit.

interestingly when the error message tripped on a clip, that clip showed no bitrate information in quicktime.

therefore, for me, 540 40 is no bitrate improvement, just more errors

http://www.huwjenkins.com/108mbit.jpg
http://www.huwjenkins.com/57mbit.jpg

mpgxsvcd
06-28-2010, 06:30 AM
mike,

my calculations say for 545, you are writing about 140mbits/sec so you need a table setting of above 32, but then again, i was never very good at maths at school lol

today, i will try all t settings set to 35 with 545 and let you know

huw

What is the equation to calculate Q settings vs. actual recorded bit rate?

sparedog
06-28-2010, 07:50 AM
i am working from my own equation, nothing definate yet. i'll explain

my table tests said the following

all quality set to 100 - max bitrate recorded 22.33
all quality set to 200 - max bitrate recorded 49.68
all quality set to 400 - max bitrate recorded 106.60

so far, in my tests, table settings have had no effect on top bitrate, just the difference between top bitrate and bottom bitrate, with lower table settings giving smaller differences.

hope this helps

mpgxsvcd
06-28-2010, 08:48 AM
i am working from my own equation, nothing definate yet. i'll explain

my table tests said the following

all quality set to 100 - max bitrate recorded 22.33
all quality set to 200 - max bitrate recorded 49.68
all quality set to 400 - max bitrate recorded 106.60

so far, in my tests, table settings have had no effect on top bitrate, just the difference between top bitrate and bottom bitrate, with lower table settings giving smaller differences.

hope this helps

Your numbers are precisely what I have experienced so far as well.

mike797
07-01-2010, 03:40 AM
Any more luck with stable settings?
I have reverted back to my original settings of Q545,500,455,410 T24

sparedog
07-01-2010, 04:24 AM
Any more luck with stable settings?
I have reverted back to my original settings of Q545,500,455,410 T24

stability sure is prooving to be tricky lol

one thing you could try with Q545,500,455,410 is setting the table settings back to the factory ones

i'm starting to think that to get 100 percent stable settings the following has to be true:
if the top bitrate is as much as 106mbits, then the table settings have to be very high meaning the bottom bitrate is very low, even as low as 18mbits.

so far i have had some luck with very high table rates with all set to 400, but i dont think these are really that useful, because of the massive bitrate lattitude. has anyone found any thing different?

at the moment, i am really testing the hell out of 400 24, to see if it really is stable once the 'switch on-error once' has occurred. if this is the only problem with it, then that is extremely livable with!!! :2vrolijk_08:

Grunf
07-01-2010, 01:22 PM
in fact, does anyone know of any reason (moire, resolution, etc) the GF1 100mbit 422 footage would not pass broadcast standards as main camera footage, if for example i wanted to make an underwater documentary using gf1s in housings?

Unfortunately, I believe there are couple of situations when I personally wouldn't accept it.

For example:

I shot this with 81Mbit and chroma just fell apart where I had lot's of low-contrast dark-green details. All the bits went into encoding hi-frequency flower chaos and there was just no data left to encode leaves.
Granted, this is probably worst-case scenario (lot's of chroma-intensive random detail, hi contrast) . AVCHD fared better (despite 1080p and roughly half bitrate). Using 4:2:2 would probably make things even worse and push the bitrate above upper limit.

I still believe that MJPEG should be used as special purposes one-trick pony only (low-light, chroma-keying, soft-focus etc.) AVCHD is just plain better as soon as there is lot's of details in the image. 30p is quite unusable in PAL-countries as well and 25p just frame-skips.

Settings:
MJPEG E1 Q: 280
MJPEG E1 T: 14
MJPEG E2 Q: 226
MJPEG E2 T: 14
MJPEG E3 Q: 200
MJPEG E3 T: 14
MJPEG E4 Q: 170
MJPEG E4 T: 14

4:2:2 unchecked.

Average bitrate: 81Mbps

http://i1024.photobucket.com/albums/y306/grunf12/81Mbit_MJPEG_frame.jpg

P. Harrill
07-01-2010, 01:26 PM
On the upside, that looks like something Monet would have painted. Of course, that's the downside, too.

mpgxsvcd
07-01-2010, 02:06 PM
I shot this with 81Mbit and chroma just fell apart where I had lot's of low-contrast dark-green details. All the bits went into encoding hi-frequency flower chaos and there was just no data left to encode leaves.
It's really worst-case scenario (lot's of chroma-intensive random detail, hi contrast) . AVCHD fared better (despite 1080p and roughly half bitrate). Using 4:2:2 would probably make things even and push the bitrate above upper limit.

I still believe that MJPEG should be used for special purposes only (low-light, chroma-keying, soft-focus etc.) AVCHD is just plain better as soon as there is lot's of details in the image. 30p is quite unusable in PAL-countries as well and 25p just frame-skips.

Settings:
MJPEG E1 Q: 280
MJPEG E1 T: 14
MJPEG E2 Q: 226
MJPEG E2 T: 14
MJPEG E3 Q: 200
MJPEG E3 T: 14
MJPEG E4 Q: 170
MJPEG E4 T: 14

4:2:2 unchecked.

Average bitrate: 81Mbps



Do you have a sample picture from the AVC-HD file? Even your MJPEG sample looks better than just about any other camera video I have seen though.

Grunf
07-01-2010, 02:09 PM
Do you have a sample picture from the AVC-HD file? Even your MJPEG sample looks better than just about any other camera video I have seen though.


Left: AVCHD @ 35Mbit sec, 1080p25 scaled down to 1280x720 (we have to compare apples to apples)
Right: MJPEG @ 81Mbit sec, 1280x720 native.

http://i1024.photobucket.com/albums/y306/grunf12/avchdvsmjpeg.jpg

mpgxsvcd
07-01-2010, 03:33 PM
Left: AVCHD @ 35Mbit sec, 1080p25 scaled down to 1280x720 (we have to compare apples to apples)
Right: MJPEG @ 81Mbit sec, 1280x720 native.

http://i1024.photobucket.com/albums/y306/grunf12/avchdvsmjpeg.jpg

Excellent comparison. I just hope that they get the 1080p @ 24 fps working for the GF1 soon.

sparedog
07-02-2010, 02:31 PM
I shot this with 81Mbit and chroma just fell apart where I had lot's of low-contrast dark-green details. All the bits went into encoding hi-frequency flower chaos and there was just no data left to encode leaves.
Granted, this is probably worst-case scenario (lot's of chroma-intensive random detail, hi contrast) . AVCHD fared better (despite 1080p and roughly half bitrate). Using 4:2:2 would probably make things even worse and push the bitrate above upper limit.


4:2:2 unchecked.

Average bitrate: 81Mbps



it is difficult to compare the two codecs, without and with 4:2:2 as without 4:2:2, the color will fall apart massively, hence why we are keen to use 4:2:2 :)

Grunf
07-02-2010, 03:58 PM
it is difficult to compare the two codecs, without and with 4:2:2 as without 4:2:2, the color will fall apart massively, hence why we are keen to use 4:2:2 :)

It this kind of situations, 4:2:2 would just make thing worse. There would be even less bits available to encode luma details.

zcream
07-03-2010, 05:29 AM
Grunf. How did you manage to get 1080p25 on AVCHD in your GF1 ? Mine only does 720p25
What settings do you use ?

Grunf
07-03-2010, 07:52 AM
Grunf. How did you manage to get 1080p25 on AVCHD in your GF1 ? Mine only does 720p25
What settings do you use ?

Are you sure you have enabled "FHD" and "AVCHD" in camera and Native "24p/25p" in Firmware patch? What's preventing you to get 1080p? Do you get card write errors?

mimirsan
07-03-2010, 10:11 AM
Are you sure you have enabled "FHD" and "AVCHD" in camera and Native "24p/25p" in Firmware patch? What's preventing you to get 1080p? Do you get card write errors?

There is no 1080p AVCHD on GF1...GH1 Yes...GF1 no...not yet anyways :-)

Grunf
07-03-2010, 12:09 PM
There is no 1080p AVCHD on GF1...GH1 Yes...GF1 no...not yet anyways :-)


Oh, I'm sorry. Everything I wrote is about GH1, not GF1. My bad.:-Dum(DBG):

canuck88
07-16-2010, 12:03 PM
I have a GF1 in the mail on it's way... If I just want the best 720p output (not interested in 1080 as my 5Dmk2 does that just fine), what settings should I be doing on the GF1? I often color grade in magic bullet looks. Thanks!

sparedog
07-18-2010, 12:35 PM
canuck88, it seems that you can use anyone of the higher bitrate settings (adammada's, mike797, papasarts, etc), as they all will record fine once your have tripped the write error, so long as you have a class 10 card

my current settings are '400 10'
follow the settings on my vimeo link below but set all q to 400 and all t to 10
http://www.vimeo.com/12918955