Improving detalization in hacked GF1

P

Pipis213

Member
Hello All!

I'm a happy owner of a hackable GF1 + its marvellous 20 pancake. I use a hacked firmware with it with dramatically increased bit-rate, removing pull-down and etc. It looks fine with movements (after finishing in Vegas) but I still don't understand the one thing - how can I improve the quality of the picture?
Here is a sample video just from the camera in MJepeg 30 frames per second (720p):
http://www.youtube.com/watch?v=8p-hoF6oZQU
From my point of view there is enough light in the scene. I used manual mode with fixed exposure and aperture, also ISO was locked at 100 value.
From my point of view the detalization of the scene is not as good as it can be. What is a clue for improving picture detalization in this camera?
The video looks a little bit blurry and I can't see small details in it.

The used setting in the firmware are given below:
[Information]
Comment=Mine
SD_Card=Any
Camera=GF1 v1.22
[Settings]
Version increment=1
Prevent version compare=Checked
Third-party battery=Checked
30min limit removal=Checked
PAL<->NTSC Menu=Checked
Manual Movie Modes=Checked
720p30 420->422 color sampling=Checked
E1 Quality=384
E1 Table=24
E2 Quality=330
E2 Table=24
E3 Quality=300
E3 Table=24
E4 Quality=276
E4 Table=24
720p59,94fps->29,97fps=Checked
720p50,00fps->25,00fps=Checked
Video Bitrate FHD/SH=50000000
Video Bitrate H=50000000
Video Bitrate L=32000000
Overall Bitrate=56000000
720p50 GOP Size=3
720p60 GOP Size=3
 
a couple of things,
what was your f stop for this shot?
maybe you taxing the bitrate. what was the file's bitrate?

you are shooting handheld on an incredibly complex scene, so maybe on a tripod you would see more detail.
there is light, but the light is very soft and not light which defines objects (like sunlight would do)
 
As to f-stop - I used 1.7 as aperture and 1/30 as shutter (it is impossible to do longer exposure in GF1).
As to the bitrate - it is 73950 kbps with 30 full frames per second, resolution 1280x720.

Yes, I did it hand-held. During my experiments I shoot some clips using my tripod - with no improving of the detalization.

Yes, the light is rather dull but we won't have sun for several weeks from now. So it is the only available source of light here now. Do you think it is because of poor lighting?
 
I am interested in this thread because I also would like to improve quality both with MJPEG and AVCHD.

I have a question and some comments.

The clip does look a little dark/flat. In direct sunlight the clip would have much more shadows = detail = data.

An Overall Bitrate=56000000 bits converts to an overall of 6,835.9375 kilobytes or 54687.5 kilobits.?.? How did you reach 73950 kbps?

Did you try upping the shutter to 1/60? But the lighting conditions might not be possible at 1/60.

I have never shot below 1/60 before. I would rather up the ISO. At a higher ISO, you should be able to up the shutter.

@sparedog, does the f-stop make a huge difference?
 
Last edited:
sxm said:
I am interested in this thread because I also would like to improve quality both with MJPEG and AVCHD.

I have a question and some comments.

The clip does look a little dark/flat. In direct sunlight the clip would have much more shadows = detail = data.

An Overall Bitrate=56000000 bits converts to an overall of 6,835.9375 kilobytes or 54687.5 kilobits.?.? How did you reach 73950 kbps?

Did you try upping the shutter to 1/60? But the lighting conditions might not be possible at 1/60.

I have never shot below 1/60 before. I would rather up the ISO. At a higher ISO, you should be able to up the shutter.

@sparedog, does the f-stop make a huge difference?
Click to expand...

Yes, the scene is flat but it is because of the lighting. Very dull and you can't do anything with that except post processing. For example in Vegas I can increase contrast of the clip and change its light balance a little bit. After that it looks like much better.
I didn't try to increase the shutter speed. Theoretically it may help in a case of a hand-held camera. I'll try to try the same scene with the same lighting but using a tripod and changed exposure.

Absolutely don't have an idea with the numbers. Media info and windows reports these values :)

Also the scene looks a little bit blurry, no small details are observable. May be USM sharpening will help?
 
Last edited:
I've tried to do the following:
1. Shot in 1/30; 1/50; 1/80 shutter with 30p - the result is identical;
2. Shot in 1/30; 1/80 shutter with 60p - the result is the same.

So the frame rate (which in GF1 is always 30p or 25p even if we shoot in 60p or 25p because the last formats just doubles inside of the camera) doesn't take any effect on the quality. The shutter is also provides the same result. So the picture is still dirty a little bit. I don't see small details.
 
If you do 3 things differently from now on, you'll get better GF1 video.

1. If you're shooting NTSC 30p, do the PTool hack to get manual shutter control and always shoot at 1/60. Period.

2. Stop waving the camera around. If you must pan, pan much more slowly. The best camera in the world looks like crap when waved around like the sample.

3. Don't expect raw video out of your GF1 to look like a Hollywood movie, TV commercial, or the Vimeo clips shot by pros like Phil Bloom etc. Nothing comes out of any camera looking that good. Shooting clean footage is only half the battle -- color correction, contrast grading, and sharpening effects are what separates professional looking DSLR video from what you posted above. Video editors like Final Cut, Adobe Premiere, Sony Vegas, and even Apple iMovie have post-processing tools, while advanced plugins like Magic Bullet Looks, Mojo, and Colorista take this to the next level.

But really, just slow down your camera movement, hack your GF1 to enable shooting with 180 degree shutter (1/60 for 30p, 1/50 for 25p), and you should see a marked improvement in your video.


Pipis213 said:
I've tried to do the following:
1. Shot in 1/30; 1/50; 1/80 shutter with 30p - the result is identical;
2. Shot in 1/30; 1/80 shutter with 60p - the result is the same.

So the frame rate (which in GF1 is always 30p or 25p even if we shoot in 60p or 25p because the last formats just doubles inside of the camera) doesn't take any effect on the quality. The shutter is also provides the same result. So the picture is still dirty a little bit. I don't see small details.
Click to expand...
 
shaveblog said:
If you do 3 things differently from now on, you'll get better GF1 video.

1. If you're shooting NTSC 30p, do the PTool hack to get manual shutter control and always shoot at 1/60. Period.

2. Stop waving the camera around. If you must pan, pan much more slowly. The best camera in the world looks like crap when waved around like the sample.

3. Don't expect raw video out of your GF1 to look like a Hollywood movie, TV commercial, or the Vimeo clips shot by pros like Phil Bloom etc. Nothing comes out of any camera looking that good. Shooting clean footage is only half the battle -- color correction, contrast grading, and sharpening effects are what separates professional looking DSLR video from what you posted above. Video editors like Final Cut, Adobe Premiere, Sony Vegas, and even Apple iMovie have post-processing tools, while advanced plugins like Magic Bullet Looks, Mojo, and Colorista take this to the next level.

But really, just slow down your camera movement, hack your GF1 to enable shooting with 180 degree shutter (1/60 for 30p, 1/50 for 25p), and you should see a marked improvement in your video.
Click to expand...

I'm able Vegas user so I can improve quality of the video. But I won't be able to add details to the video if there are not any ;)

What is the purpose of setting 1/60 for 30p? I don't really understand the process of packing exposures into frames.

PS. The thread was not about moving the camera but about the details........
 
You are getting a lot of motion blur at 1/30 and the details are being smoothed when you pan. Try a higher fstop and keep the 1/50 or 1/60.
 
mojo43 said:
You are getting a lot of motion blur at 1/30 and the details are being smoothed when you pan. Try a higher fstop and keep the 1/50 or 1/60.
Click to expand...
I was asking not about blur because of the motion. I asked about detalization of the an almost still scene with mjpeg :)
This is the same scene but I used a tripod and it is brighter a little bit, this is a RAW AVCHD file http://files.mail.ru/ZXQSBK

But why should I use 1/50 or 1/60 exposure? Will it dramatically help? What is the correlation between the frame rate and the exposure?
 
Okay, one more try since it's a holiday.

Waving your camera around like that, you're never going to get a very good image. Slow down your camera movement, let your codec (video compression) breathe easier and preserve more detail.

180 degree shutter explanation here. Put simply, shoot at double the shutter speed for a given framerate. So for 30p video, lock your shutter at 1/60. For 25p, lock it to 1/50, or as close as you can since some cameras don't offer exactly 1/50 shutter speed.

mojo43 is right, your problem is you shot the sample at too low a shutter speed and with the lens wide open. That's a sure recipe for soft, blurry video. I would add that even if you'd stopped the lens down and shot at 180-degree shutter, you'd still wind up with unsatisfying video simply because you're panning too quickly. Settle down, go for smoother and slower camera movement and you'll be happier with the image.




Pipis213 said:
I'm able Vegas user so I can improve quality of the video. But I won't be able to add details to the video if there are not any ;)

What is the purpose of setting 1/60 for 30p? I don't really understand the process of packing exposures into frames.

PS. The thread was not about moving the camera but about the details........
Click to expand...
 
shaveblog said:
Okay, one more try since it's a holiday.

Waving your camera around like that, you're never going to get a very good image. Slow down your camera movement, let your codec (video compression) breathe easier and preserve more detail.

180 degree shutter explanation here. Put simply, shoot at double the shutter speed for a given framerate. So for 30p video, lock your shutter at 1/60. For 25p, lock it to 1/50, or as close as you can since some cameras don't offer exactly 1/50 shutter speed.

mojo43 is right, your problem is you shot the sample at too low a shutter speed and with the lens wide open. That's a sure recipe for soft, blurry video. I would add that even if you'd stopped the lens down and shot at 180-degree shutter, you'd still wind up with unsatisfying video simply because you're panning too quickly. Settle down, go for smoother and slower camera movement and you'll be happier with the image.
Click to expand...
Thank you very much for the 180 shutter explanation. I'll try to use it and will share the result.

PS. I'm not speaking about the pans quality, only about static scenes. Fast pans or fast moves of huge objects in a clip is a separate theme.
 
Pipis213 said:
Thank you very much for the 180 shutter explanation. I'll try to use it and will share the result.

PS. I'm not speaking about the pans quality, only about static scenes. Fast pans or fast moves of huge objects in a clip is a separate theme.
Click to expand...

I downloaded the clip. Even when it was on a tripod it was still moving... Nonetheless it wasn't very sharp either, but I think you said that you had the aperture wide open? This will bring down detail a lot since a lens's sweet spot is usually in a higher f stop so try that and choose a focus point. I can't tell what your focus point is on this shot.
 
I've executed the recommendations above:
1. Step down the aperture to 8 (from 1.7);
2. Used 1/60 exposure for 30p clip;
3. Used 100% steady tripod (actually it is a screw-clamp attached to a thick metal handrail of a balcony).

The weather is a little bit more sunny so I could use fast shutter.

The result in raw is here http://files.mail.ru/BHRYUW

To be honest - I can't say that the video has more details as the video with 1/30 exposure and 1.7 aperture. Any comments?

PS. Also I'm preparing a mix of videos with 1/30-1/500 apertures. Will publish it when finish it.

Aha, I've compared the same scene with 100% stable tripod and with very narrow aperture and different shutter speeds. The result is here http://www.youtube.com/watch?v=QhG4VgVODdI
It is obvious that all the slices are the same in the details. Only the falling snow is different at every piece of the video. Any other suggestions? Aperture also was shrunk. No difference.
 
Last edited:
i am not quite sure what you are expecting, but i have just watched the new video and that is pretty much what you will get out of 720p video. it looks good to me.

video should never be compared to a photograph. the intention of video is not 'oh wow, look how sharp it is!'. it is a story telling device, invented for television to tell stories and not to be analysed for sharpness. this is why for many years people were happy with standard definition.

with this in mind, different shutter speeds wont help video, they just make it look strange, as nothing made on television is made without 180 degrees shutter angle. yes in feature films occasionally, ridley scott or steven speilberg and other directors will use a higher shutter speed, but that is to create a phsycolical effect on the viewer, telling the viewer that this scene is not a normal occasion. but when you watch a documentary, or a soap opera, or talk show it is shot with a 180 degree shutter angle.

the shots you shoot are interesting, because i do not live in russia, so it is interesting to see how russian apartments buildings are. for me, added detail would not enhance the shot, i know what a tree is, so i do not need to see details in the branches :D
 
sparedog said:
i am not quite sure what you are expecting, but i have just watched the new video and that is pretty much what you will get out of 720p video. it looks good to me.
Click to expand...
Thank you sparedog for your comment.
Yes, this was a common view of Moscow and other Russian huge cities.

The video looks like small details are blown from the frames, or may be tones variations are posterized. It is not only matter of local contrast, may be not the problem of the local contrast. To be honest I expected much better result from the camera. I hope that in close scenes the result will be good.
I don't understand what is the reason of such behaviour - noise reduction mechanism, codec, small matrix, bad lenses?

From my point of view the lenses (20mm) are superior and it is ahead of a great number of other lenses. The video looks like it is just unscaled from much weaker resolution. The next week I'm going to test the camera in a studio with controlled lights, and short and middle light scenes.
 
maybe it is camera specific?
i have had footage i shot at 100mbits recently tested by bbc and getty footage and it has passed all broadcast standards, for both television and commecial use. in their tests, it surpassed all other dlsr footage, and was considered better than 1080 shot on dlsrs due to the codec limitations other dlsrs have. i only say this, because maybe the problems you see are specific to your camera?
 
I agree that the 180 degree shutter rule is important. However, I don't agree that it should always be adhered to. There are many situations where 1/60th of a second or greater shutter speed will get you nothing but a back screen. In those cases 1/30th of a second shutter speed can work if you get everything else right.

Check out the first clip in the video below. I used a 55mm F1.2 lens with a 1/30th of a second shutter speed just to be able to see anything at all under that dark overhang. However, I think the slow motion portion turned out fairly well considering the conditions. If I had gone with 1/60th of a second noise or dark video would have been an issue. I think I had the ISO around 640-800. It only goes to 1600 in video on the GF1 and it looks terrible when you use ISO 1600.

 
sparedog said:
maybe it is camera specific?
i have had footage i shot at 100mbits recently tested by bbc and getty footage and it has passed all broadcast standards, for both television and commecial use. in their tests, it surpassed all other dlsr footage, and was considered better than 1080 shot on dlsrs due to the codec limitations other dlsrs have. i only say this, because maybe the problems you see are specific to your camera?
Click to expand...
To be sure that it is a camera specific it is good to test two and more cameras together shooting the same scene. I don't have anyone next to me with the same camera. I'll try to execute a short studio test of the camera and share the results.
 
You must log in or register to reply here.
Back
Top