GF1 FHD AVCHD Menu Option

S

spedcake123

New member
After downloading the latest version of Ptools, I noticed a FHD option under patches for testers>middle level research>menu related. So I ticked the option and loaded the hacked firmware onto my GF1 to test it out.

Using the star chart, I took three clips. Average bitrate using my settings was 48.2 Mbps [max bitrate I got was 50.2 Mbps]. I got a write speed error for the second clip but not for the first and third. However, all of them had a black transparent bar for the bottom half of the video.

Click for full size - Uploaded with plasq's Skitch


I then took the same shot in both SH and FHD modes using Manual mode, and my observations were that the FHD file took approximately 4 times the size of the SH file and looked more sharper. Using mediainfo, I spotted that the FHD file was interlaced, while the SH file remained progressive.
720p (SH):

Click for full size - Uploaded with plasq's Skitch

1080p (FHD):

Click for full size - Uploaded with plasq's Skitch

My current settings for AVCHD:
FHD/SH: 60000000
H: 60000000
L: 60000000
Overall: 70000000

I'm using a Sandisk 8GB Class 6 card. I'll do some daylight testing when I next get the chance :)
 
wowah, you're not having any of those (VHS gone wrong) while recording? I'm using class Class 10, and everyone says that FHD was included just for fun, but you were able to with a class 6?! :o How'd you do mate?!
 
Try this:
Set the camera to PAL mode using the new firmware menu option.
Set the camera to record 1080p AVC-HD with full manual controls.
Set the shutter speed to 1/50 of a second.
Now record the video and see what happens. Keep recording separate clips and make the shutter speed faster each time. My testing showed that any faster than 1/60 of a second will produce the static band across the top. The band gets bigger the faster the shutter speed with the PAL format selected.

The opposite happens with NTSC. You start getting the band any slower than about 1/400 of a second. It will cover the whole video with anything less than 1/120 of a second.

Here is a PAL sample of the 1080p video from the GF1. It is not as good as the 720p MJPEG though.
http://vimeo.com/13597744

Shooting video between 1/30 and 1/60 is much more realistic so right now the PAL 1080p option is a better one. However, it still produces interlaced video which I hate so I gave up on it. I hope that they figure out how to get native 1080p on the GF1. I won’t buy another camera if they do.
 
Who said that FHD is just for fun??
My fhd 'C' settings work without any errors on kingston class 4 card...
 
mpgxsvcd said:
Shooting video between 1/30 and 1/60 is much more realistic so right now the PAL 1080p option is a better one. However, it still produces interlaced video which I hate so I gave up on it. I hope that they figure out how to get native 1080p on the GF1. I won’t buy another camera if they do.
Click to expand...
Wait, so the GF13 now shoots honest to god 1080p (not upresed one like in MJPEG mode)? Did you do your resolution test?

Perhaps 24p/25p native patch is next candidate for porting then ...
 
svecher said:
Wait, so the GF13 now shoots honest to god 1080p (not upresed one like in MJPEG mode)? Did you do your resolution test?

Perhaps 24p/25p native patch is next candidate for porting then ...
Click to expand...

I am doing the resolution test now. It is difficult to tell though. The GF1 currently records 1080i @ 29.97 or @ 25 FPS. It is not like the GH1 that records 1080p @ 24 FPS embedded in a 1080i container. With the GF1 you can not extract the 1080p @ 29.97 video without resolution loss.

What we need is to get the GF1 to record native 1080p @ 24 FPS. Then I actually think it will match the GH1.
 
mpgxsvcd said:
Vitaliy said it is just for fun. Try different shutter speeds for FHD GF1 video and you will see the errors.
Click to expand...
I have a GH13, so this is a matter of curiosity for me.

mpgxsvcd said:
I am doing the resolution test now. It is difficult to tell though. The GF1 currently records 1080i @ 29.97 or @ 25 FPS. It is not like the GH1 that records 1080p @ 24 FPS embedded in a 1080i container. With the GF1 you can not extract the 1080p @ 29.97 video without resolution loss.
Click to expand...
I have doubts on that last sentence. Resolution loss is usually associated with de-interlacing footage from cameras that shoot in interlaced mode natively. GH1 natively shoots progressive footage and wraps it into interlaced wrapper (see here for the mechanics of the process
http://www.dvxuser.com/V6/showpost.php?p=2052273&postcount=312). I also doubt that GF13 can shoot 1080p @ 29.97p. FWIW, most media programs will report 29.97 fps frame rate (60i) on 1080/24PsF footage shot on GH1, however the underlying progressive frame-rate is 24 fps, once 2:3 pulldown is removed. VK can confirm exactly what's going on, but in the meanwhile you can try to perform reverse pulldown as described for GH1 in workflows section. Also, some NELs, like Adobe CS5 are reportedly capable of performing RP on the fly and dropping 24PsF directly into 24p timeline. Or post a sample and I'll analyze it.
 
svecher said:
I have a GH13, so this is a matter of curiosity for me.


I have doubts on that last sentence. Resolution loss is usually associated with de-interlacing footage from cameras that shoot in interlaced mode natively. GH1 natively shoots progressive footage and wraps it into interlaced wrapper (see here for the mechanics of the process
http://www.dvxuser.com/V6/showpost.php?p=2052273&postcount=312). I also doubt that GF13 can shoot 1080p @ 29.97p. FWIW, most media programs will report 29.97 fps frame rate (60i) on 1080/24PsF footage shot on GH1, however the underlying progressive frame-rate is 24 fps, once 2:3 pulldown is removed. VK can confirm exactly what's going on, but in the meanwhile you can try to perform reverse pulldown as described for GH1 in workflows section. Also, some NELs, like Adobe CS5 are reportedly capable of performing RP on the fly and dropping 24PsF directly into 24p timeline. Or post a sample and I'll analyze it.
Click to expand...

Yes I am not sure what is going on with it. My theory is that it is 1080i @ 29.97 FPS. However, I will post a sample later on today that hopefully you guys could analyze and determine what it is.

I shot clips for 1080p in both PAL and NTSC just now.
 
I shot some PAL footage today. 48 mbps. Nice! Is it just me or does this footage not play back in camera ?
 
zcream said:
I shot some PAL footage today. 48 mbps. Nice! Is it just me or does this footage not play back in camera ?
Click to expand...

That is right it will not playback in camera. I have compared the 1080i footage out of the GF1 to the high bit rate 720p MJPEG and the 1080i isn't even close. I really am fairly satisfied with 720p MJPEG.

I am almost hoping the GH2 has 1080p MJPEG instead of 1080p AVC-HD. Memory cards are cheap. What I really want is 100mb/sec 1080p @ 24 FPS 4:2:2. I would quit fiddling and start shooting more if we get that.

Has anyone tried the 1080p MJPEG on the GH1 to see if it is any better than 720p MJPEG? I have access to a GH1 but the owner won't hack it yet so I can't test it.
 
To my eyes, the 1080i AVCHD (once the 25P is extracted) is sharper resolution wise than 720p MJPEG.
How do you find it better ?

mpgxsvcd said:
That is right it will not playback in camera. I have compared the 1080i footage out of the GF1 to the high bit rate 720p MJPEG and the 1080i isn't even close. I really am fairly satisfied with 720p MJPEG.

I am almost hoping the GH2 has 1080p MJPEG instead of 1080p AVC-HD. Memory cards are cheap. What I really want is 100mb/sec 1080p @ 24 FPS 4:2:2. I would quit fiddling and start shooting more if we get that.

Has anyone tried the 1080p MJPEG on the GH1 to see if it is any better than 720p MJPEG? I have access to a GH1 but the owner won't hack it yet so I can't test it.
Click to expand...
 
EDIT:- I looked at your images. Look at the trees on the right side. Click 720p and then 1080p. There is clearly more detail in those trees in 1080p.
Look also at the rooftop on the right. Clearly sharper in 1080p.
How did you decide from these pics that 720p is sharper ?

>>
How did you deinterlace ? Its basically 25p in a 50i container. Deinterlacing is done to actual interlaced footage to make it progressive.
I have to do tests, but on first observation, even if the footage is 720p upscaled, it should not be "much less sharp".
 
Last edited:
svecher said:
Why do both screenshots have resolution of 1675x938 px ?
Click to expand...

Oh, that's how it uploaded then, forgot to check. You could also say it was like a resolution test as in theory, 1080p downscaled should still be better than 720p upscaled but it isn't. I'll post up the proper resolution samples if anyone wants me to.


zcream said:
EDIT:- I looked at your images. Look at the trees on the right side. Click 720p and then 1080p. There is clearly more detail in those trees in 1080p.
Look also at the rooftop on the right. Clearly sharper in 1080p.
How did you decide from these pics that 720p is sharper ?
Click to expand...

If you look at the street lamp and the post box, you'll see that 720p is sharper and crisper and to my eyes, the trees and rooftops seem to have more detail in 720p. I guess it differs between everyone, but that was my opinion.
 
Last edited:
spedcake123 said:
Oh, that's how it uploaded then, forgot to check. You could also say it was like a resolution test as in theory, 1080p downscaled should still be better than 720p upscaled but it isn't. I'll post up the proper resolution samples if anyone wants me to.
Click to expand...
The problem is that unknown upscaling/downscaling mechanisms are thrown into the mix. When making such comparisons, it is best to review samples at their native resolution, so please repost. Thanks.
 
The street lamp looks terrible in 1080p. Not only that, there is purple fringing and what seems similar to moire. But that does not seem to be resolution, but a compression artifact. Resolution will not introduce those jaggy lines. It will only make things softer.
Are you sure you are extracting the 24p/25p properly ?
Remember also the gotchas with the 1080p. It really ONLY works properly in 1080p/25P and 1/50th.
 
You must log in or register to reply here.
Back
Top