modified GF1 at 100mbits 4:2:2 for underwater?

a bunch of us over on dvxuser have been getting some amazing results with the modified gf1 and 100mbit 4:2:2 720 video. with it's small body size and a relatively large sensor size, i am thinking this would make a great underwater setup.

is there anything that would stop video at 100mbits and 4:2:2 from being broadcast-able for nat geo channel, discovery etc, if i apply drop frame (30fps->29.97fps)in post? how can i test for moire, etc to see minimum broadcast standards?

the gf1 suffers from minimal 'jello' in mjpeg, uses intra frame coding and is the equivalent to a 16mm film camera when it comes to crop factor.

some test videos can be seen at

http://www.vimeo.com/12918955

I first mentioned the TESTER 13 hack (for Panasonic GH1 and GF1) a few posts back. I'll have my GF1 in a few days time and will do a full evaluation for my own interest...and pass it on here if anyone is interested. What I read so far sounds very promising.

The great thing is that it's simply a firmware hack that unlocks the parameters within the camera...no physical modification of the hardware is required.

The hack is evolving rapidly so it may not be too long before the GF1 can do true 1080p @ 4:2:2. At the moment it's upscaled 720p @4:2:0. It's hard to say how crusty institutions like the BBC will regard the 720p...even at such high bit rates and at 4:2:2. (From what I've read, the AVCHD @ 35mbps actually looks better than the mjpeg @ 100 mbps). Their main objection to the Canon 5D2 and 7D footage was the low video resolution and excessive moire/aliasing...because of the way Canon's line-skip when reading the sensor.

The Panasonic GH1/GF1 seems to have a major advantage in this regard because 1) the sensor is a lot smaller; and 2) the algorithm seems to do a better job of reading the sensor for HD video purposes (no one seems to know if it line skips or not). The video resolution and lack of aliasing seems to be a lot better than the Canons. You can see this in a good comparison:

Panasonic are about to announce a GH2...the rumours are flying around that it will have a Global shutter, which would eliminate Jello and aliasing in one stroke. I'd be surprised if it did, but no doubt Panasonic will somehow react to the hack ( the way Canon did to Magic Lantern) and bring out a less intentionally crippled camera than they may have otherwise.

If it indeed does have a Global shutter, and if the hack can work with it's firmware to give very high bitrate...it could just be the Scarlet alternative. At least until JJ sorts out the double whammy that just hit Scarlet's production process.

Just an aside...if anyone has used an AG LA7200 anamorphic lens underwater please PM me. I want to shoot anamorphic 2.35:1 with my GF1, but I'm curious how this would work behind a large dome.

The Panasonic GH1/GF1 seems to have a major advantage in this regard because 1) the sensor is a lot smaller; and 2) the algorithm seems to do a better job of reading the sensor for HD video purposes (no one seems to know if it line skips or not). The video resolution and lack of aliasing seems to be a lot better than the Canons. You can see this in a good comparison:

the more reliable sources say that it pixel drops, rather like how photoshop resizes images. this means it doesn't introduce too many noticeable artifacts.

the more reliable sources say that it pixel drops, rather like how photoshop resizes images. this means it doesn't introduce too many noticeable artifacts.

I'm inclined to agree with you, but I've also been amused by some of the speculation (even regarding Canon) about this on the various forums:

Barry Green:

"It's been a source of raging debate since the 5D first hit, over a year and a half ago. The reigning theory is that it skips pixels and skips lines. The only thing we've been told by Canon is that they use a six-pixel bin, and don't skip lines. But those who have put it under the microscope reject that, saying that the results the Canon gets are absolutely in line with what one would expect if it was skipping two lines out of three.

And that skipping of rows and pixels, combined with the cheap downconversion hardware, results in massive aliasing and especially purple/orange moire.

The GH1 probably does some things similarly, but it's immune to the purple/orange aliasing. It has luma aliasing, but not chroma aliasing.

So, nobody really knows, and the only statement we have from anyone in the factory seems contradictory to the theories that have been put forth and tested. "

To which some cynic answered:

"Since the manufacturers don't brag about how they do it, I suppose they must have something to hide -- which strongly suggests the skipping scenario. "

Barry Green:

" Or they have trade secrets or patents on their technology, and therefore don't want to go handing those trade secrets to their competitors..."

As far as I'm concerened, it's the end results that matter. The hacked GH1 & GF1 seem to have much better video resolution than Canon and a lot less aliasing.

__________________

It's hard to say how crusty institutions like the BBC will regard the 720p

You mean the world leaders in Natural History film-making ... It does pee some people off when they save up to buy expensive (to them) systems only to find Broadcast companies are not knocking at their door and begging to buy their footage or fund their films. They obviously have their reasons

What about Discovery or Nat Geo maybe their not so crusty in their attitudes

Dive safe

DeanB

You mean the world leaders in Natural History film-making ... It does pee some people off when they save up to buy expensive (to them) systems only to find Broadcast companies are not knocking at their door and begging to buy their footage or fund their films. They obviously have their reasons

 

What about Discovery or Nat Geo maybe their not so crusty in their attitudes

 

Dive safe

 

DeanB

here's my thinking:

the gf1 currently only shoots 30p (not 29.97) when recording in mjpeg at 4:2:2.

therefore for broadcast, i think that a simple frame remap in cinema tools will make the footage 29.97 and not have any detrimental effect on the footage.

however if the footage is intended for bbc or pal, a remap to 25 is too much of a speed difference to go unnoticed (however it doesn't half look nice when we play wildlife slightly slow lol) .

what are the bbc/discovery co-productions shot/delivered at in the uk?

a couple years ago, alot of stuff for discovery, bbc, nat geo was shot on a varicam at 720p and upres-ed in post, is this still the case? if so, 100mbits 4:2:2 is the codec specs for the varicam's dvcpro hd, so i'm thinking if the cmos chip is the gf1 is ok, then it should be perfectly broadcastable, right?

"a couple years ago, alot of stuff for discovery, bbc, nat geo was shot on a varicam at 720p and upres-ed in post, is this still the case? if so, 100mbits 4:2:2 is the codec specs for the varicam's dvcpro hd, so i'm thinking if the cmos chip is the gf1 is ok, then it should be perfectly broadcastable, right?"

Of course it's perfectly broadcastable...I've had HDV from a Sony HC1 broadcast on 1080 HDTV in Australia. But it seems that often it's not quite as simple as that. How easily the delivery format fits into the given production's post workflow can be viewed differently by different producers/editors. Some might be willing to give a different acquisition format a try (particularly if it's useful to them), others might not. One advantage with underwater material is that if its well shot and well lit, converting frame rates down (30p to 25p) is no big deal with the right software...no audio sync to worry about.

As with many new things, production/broadcast acceptance will also to some extent depend upon the personal bias and "tech politics" of the individuals involved.

Personally, I'd rather see beautiful material shot on a GF1 or a Canon 5D2 than some of the mediocre stuff I recently sat through at an IMAX theater.

"a couple years ago, alot of stuff for discovery, bbc, nat geo was shot on a varicam at 720p and upres-ed in post, is this still the case? if so, 100mbits 4:2:2 is the codec specs for the varicam's dvcpro hd, so i'm thinking if the cmos chip is the gf1 is ok, then it should be perfectly broadcastable, right?"

 

 

Of course it's perfectly broadcastable...I've had HDV from a Sony HC1 broadcast on 1080 HDTV in Australia. But it seems that often it's not quite as simple as that. How easily the delivery format fits into the given production's post workflow can be viewed differently by different producers/editors. Some might be willing to give a different acquisition format a try (particularly if it's useful to them), others might not. One advantage with underwater material is that if its well shot and well lit, converting frame rates down (30p to 25p) is no big deal with the right software...no audio sync to worry about.

 

As with many new things, production/broadcast acceptance will also to some extent depend upon the personal bias and "tech politics" of the individuals involved.

 

Personally, I'd rather see beautiful material shot on a GF1 or a Canon 5D2 than some of the mediocre stuff I recently sat through at an IMAX theater.

If you send in clips to a programme or supply footage as part of a programme then the formats are more open. I've scene sequences shot on Hi8 featured on the Beeb not long ago. Its when you approach them with a treatment or idea for a film things change. I was told not long back that 50Mbps and 4:2:2 was the golden ticket however since then I have found out its not just what the camera that can do, there are other factors.

If someone has a camera that's all singing and all dancing and they still cannot get the 'golden ticket' then they have to ask themselves WHY? there has to be another reason. There are hundreds of peeps out there with the latest and greatest which are really nothing but expensive holiday camera's... Which is, to be honest, great if that's all you want them for

Dive safe

DeanB

Anyone thinking of getting a new GF1 of GH1 to work @ high bit rates with the hack beware!

Panasonic have now locked the firmware at the factory for the latest production batches. It seems that unlike Canon, who responded to Magic Lantern by updating their own firmware to give customers what they asked for, Panasonic is desperately trying to keep its cameras in their original retarded form.

That's pretty interesting... 2nd hand ones will be more valuable than new one.... that would be a first in electronics! lol

That's pretty interesting... 2nd hand ones will be more valuable than new one.... that would be a first in electronics! lol

I think you might be right.

The supply of hackable new ones seems to be going very quickly (waiting for my two GH1's from the US......the hack also unlocks the Pal/NTSC restriction, giving users both options).

There is now many people who have done comparative tests with their own EX1/3's; 5D2's; etc and the consensus seems to be that the hacked GH1 is cabable of easily outperforming the Canons and matching (some say bettering) the EX3. The comparative frame grabs I've seen on other forums support this claim...but obviously, ergonomics are another matter.

Not bad for a tiny $1500 DSLR.

If I remember correctly, the Beeb chose the Panny 720p because of the 60p slow motion at AVC-I codec, which is a robust codec.

Now the GF1 may have a hack that allows 100mbps, but one has to match with memory cards that are fast enough to handle that. I can't remember precisely but the initial tests with the hack had drop out due to the card issue. As mentioned, the artifacting from the compression engine creates certain looks which would not clear the broadcaster's scrutiny. However, if you are selling clips, then it doesn't matter. If your production has the quality that a broadcaster will buy, then they will buy it. But if you are looking for them to finance you, the checklist is a lot longer than just cameras.

Now the GF1 may have a hack that allows 100mbps, but one has to match with memory cards that are fast enough to handle that. I can't remember precisely but the initial tests with the hack had drop out due to the card issue.

Yes, things are still very much a work in progress. The hack allows many variations of the AVCHD and MJPEG codec internal parameters...a lot of possible combinations. Users are currently testing quality vs stability and posting their findings on a couple of data bases. High quality, stable settings have now been achieved using (expensive) Class 10 SD cards but so far there are some limitations on in camera playback of recorded sequences, or using Native Progressive.

The great thing is that it's possible to select relatively high bit rate settings (but not the highest possible) which still allow full acquisition reliability and even in camera playback.

ok i have finally been able to take the camera underwater. here is a sample of the GF1 shooting 100mbits 4:2:2 720p, with full manual controls. the footage is shot with in camera sharpening turned off, so a little could be applied in post to make it pop.

the sample has been compressed hugely for vimeo.

http://www.vimeo.com/16587274

Looks great, I reckon that is a good example o what the Af100 will be like

Looks great, I reckon that is a good example o what the Af100 will be like

sadly the af100 is recording PH at i believe 24mbits (please correct me if this is wrong). this just cant compete with 100mbits at 4:2:2. yes the af100 is 1080, but thats even more pixels compressed into the PH. however, hooked up to an external recorder, i think the af100 will be awesome!

but for the size, this is tiny, you cant beat this. in it's housing it's smaller than a z1 or hc9. it's cheap so you can use it as a second cam, or in situations where you primary cam is too valuable. and you can stick an insanely expensive piece of glass on the front and get some spectacular quality.

sadly the af100 is recording PH at i believe 24mbits (please correct me if this is wrong). this just cant compete with 100mbits at 4:2:2. yes the af100 is 1080, but thats even more pixels compressed into the PH. however, hooked up to an external recorder, i think the af100 will be awesome!

 

but for the size, this is tiny, you cant beat this. in it's housing it's smaller than a z1 or hc9. it's cheap so you can use it as a second cam, or in situations where you primary cam is too valuable. and you can stick an insanely expensive piece of glass on the front and get some spectacular quality.

Apologies Huw meant to say... similar quality to Af100 if captured through HDSDI @ 100mbits 4:2:2

But you are totally right, the GH1 is a superstar!

but for the size, this is tiny, you cant beat this. in it's housing it's smaller than a z1 or hc9. it's cheap so you can use it as a second cam, or in situations where you primary cam is too valuable. and you can stick an insanely expensive piece of glass on the front and get some spectacular quality.

Which housing are you using for this little monster?

Which housing are you using for this little monster?

the 10bar. it's built like a tank (literally, it looks kinda world war 2) and has a clear backplate to check the 2 orings when you pressure test it.

i recommend it, there's more info on it and pics here

http://wetpixel.com/forums/index.php?showt...32112&st=20

the 10bar.

i recommend it, there's more info on it and pics here

http://wetpixel.com/forums/index.php?showt...32112&st=20

Hi Huw,

I read the thread you suggested and I have some questions:

- Do you have the latest release of the 10bar housing or the first one? I read that on the first one some controls are shared and some dials doesn't work very well.

- Do you have also the dome port for the 14-45?

- Are you using the manual video mode with the GF13 hack?

Thanks