1080p MJPEG is not 1080p - Proof inside

M

mpgxsvcd

Veteran
I have included several images below that show that the 1080p MJPEG is not any better than the scaled high bit rate 720p MJPEG on the GF1. In fact the 1080p MJPEG doesn't look as good as the high quality still picture scaled down to 720p and back up to 1080p.

All of the MJPEG images are the first frame from the MJPEG videos. All of the still images are JPEG images taken straight out of the camera.

The 1080p image straight out of the camera has much more resolution than the 1080p MJPEG image shows.

You will have to go to the dpreview site post below since the image links I posted before have expired.
http://forums.dpreview.com/forums/readflat.asp?forum=1041&message=35735436


I have also uploaded a sample video to youtube and vimeo that shows how the true 1080p image looks much better than the 1080 MJPEG images.

http://www.youtube.com/watch?v=xB2tgB4amP4

http://vimeo.com/13135705
 
Last edited:
Can someone try this test with a GH1 and the native 1080p @ 24 FPS setting? Also could you try 1080p MJPEG with a GH1 to see if the GH1 can do true 1080p MJPEG?

Here is the link to the resolution chart. Just film this about 12-36 inches from the chart printed out on plain paper. Then grab a 1920x1080 still frame from the AVC-HD file and grab the .jpg file produced with every MJPEG clip.

http://www.pentax-hack.info/firmware/pan/ISO_12233-reschart.pdf
 
I'm not convinced AVCHD 1080p is 1080p

I'm not convinced AVCHD 1080p is 1080p

Maybe they fooled us all and scaled 720p to 1080p in camera but just adjust the width of the video image.


mpgxsvcd said:
I have included several images below that show that the 1080p MJPEG is not any better than the scaled high bit rate 720p MJPEG on the GF1. In fact the 1080p MJPEG doesn't look as good as the high quality still picture scaled down to 720p and back up to 1080p.

All of the MJPEG images are the first frame from the MJPEG videos. All of the still images are JPEG images taken straight out of the camera.

The 1080p image straight out of the camera has much more resolution than the 1080p MJPEG image shows.

You will have to go to the dpreview site post below since the image links I posted before have expired.
http://forums.dpreview.com/forums/readflat.asp?forum=1041&message=35735436


I have also uploaded a sample video to youtube and vimeo that shows how the true 1080p image looks much better than the 1080 MJPEG images.

http://www.youtube.com/watch?v=xB2tgB4amP4

http://vimeo.com/13135705
Click to expand...
 
Thats some interesting results!
My 2 cents...when I upscale the 720p in the pitiful editing programs I use it looks a bit crappy...so I use 1440x1080 setting in cam...saves me a lot of bother and looks great....a lot better than the 720p...so I guess at the end of it all...its down to what works best for the individual. :-)

I have to ask
A 1080p jpeg photo is a single frame ...it would be unlikely that 1080p mjpeg video would be exactly the same image quality (being a series of jpegs one after the other right?)...isnt something lost in the image when shooting video anyway (with compression etc)?
 
Last edited:
mpgxsvcd
do you think then it's best to shoot 720p and upscale in post?

i use after effects and magic bullet instant hd which gives great results with the up-res. does anyone get even better results scaling 720to1080 with another after effects plugin?
 
sparedog said:
mpgxsvcd
do you think then it's best to shoot 720p and upscale in post?

i use after effects and magic bullet instant hd which gives great results with the up-res. does anyone get even better results scaling 720to1080 with another after effects plugin?
Click to expand...

Definitely, scale in post processing. The pictures above show that for the GF1 at least scaling in the camera is pointless.

I still would like to see what the GH1 1080p MJPEG can do with the resolution chart. It would be great if the GH1 can do true 100 mb/sec 1080p MJPEG.
 
Do GH1 and GF1 have same sensor capability?

Do GH1 and GF1 have same sensor capability?

http://panasonic.net/avc/lumix/line_up/systemcamera.html

That Panasonic page shows all their 4/3 cameras have 12.1 effective megapixels. Both GH1 and GF1 (and G2 and GH10) have 12.1 effective megapixels. What's the difference?

- GH1 uses more of its 12.1 effective megapixels than a hacked GF13 does when shooting 1080?

- Or GH1 is also always shooting 720 but, using a better up rez code than a hacked GF13 version does when saving to 1080?

Which do you think it is? (Opinions welcome. Real GH1 tests are even more welcome...)
 
TheNewDeal said:
http://panasonic.net/avc/lumix/line_up/systemcamera.html

That Panasonic page shows all their 4/3 cameras have 12.1 effective megapixels. Both GH1 and GF1 (and G2 and GH10) have 12.1 effective megapixels. What's the difference?

- GH1 uses more of its 12.1 effective megapixels than a hacked GF13 does when shooting 1080?

- Or GH1 is also always shooting 720 but, using a better up rez code than a hacked GF13 version does when saving to 1080?

Which do you think it is? (Opinions welcome. Real GH1 tests are even more welcome...)
Click to expand...

I think that the GF1 always renders the raw video input to 1280x720 before it writes it to the final format. No matter what format or resolution you write it to it will always be sourced from 720p. That is just my hypothesis though.
 
I just tested the 1080p MJPEG and it definitely looks like up scaling up from 720p... it just doesn't compare to 1080 FHD unfortunately.
 
I believe it has already been established in multiple threads that MJPEG 1080p is just 720p upscaled. The recommendation (I believe it was Barry Green) was to upscale it in post rather than having the camera do it for you. Camera scalers always...um...sucks . AVCHD however looks a lot better....sharper and more detailed (not that the MJPEG looks bad at all....it doesn't).
 
I was able to get my hands on an unhacked GH1 today. I ran the same test with it and the GF1. The 1080p AVC-HD file looks a lot more like the 1920x1080 still frame than the GF1’s video does. In fact the unhacked GH1’s 720p MJPEG video looks sharper than the hacked 100+ mb/sec GF1’s MJPEG video.

I will post the footage on Vimeo later today so you can download the original footage. I will post one file with everything basically in a lossless compressed file and then post individual files straight out of the camera.

Here are the JPG images for now.

http://www.dpreview.com/galleries/4222674355/albums/gf1-vs-gh1

Bottom line: If you are looking for the ultimate video quality for under $10K right now a hacked GH1 is probably your best option.
 
Thanks for posting these comparisons, mpgxsvcd. The GH1 clearly produces higher quality video frames than the GF1, and I think it's very likely due to a significantly improved sensor scanning algorithm in the GH1. But since the exact technique is apparently a trade secret that Panasonic hasn't yet disclosed, I'm not sure we can do much more than speculate at this point.
 
mpgxsvcd said:
I was able to get my hands on an unhacked GH1 today. I ran the same test with it and the GF1. The 1080p AVC-HD file looks a lot more like the 1920x1080 still frame than the GF1’s video does. In fact the unhacked GH1’s 720p MJPEG video looks sharper than the hacked 100+ mb/sec GF1’s MJPEG video.

I will post the footage on Vimeo later today so you can download the original footage. I will post one file with everything basically in a lossless compressed file and then post individual files straight out of the camera.

Here are the JPG images for now.

http://www.dpreview.com/galleries/4222674355/albums/gf1-vs-gh1

Bottom line: If you are looking for the ultimate video quality for under $10K right now a hacked GH1 is probably your best option.
Click to expand...

I too have both the GF1 & GH1 running the same settings (modulrs...gf1 mjpeg at 720p 4.2.2...gh1 mjpeg 1440x1080p ) theres a big difference especially in detail image quality (gf1s a fair bit softer)

Both are bargain cameras for sure!
 
I realize that the conclusions in this thread have been widely accepted for a long time. However, I had not seen any definitive evidence that showed the difference between the GF1 video resolution and the GH1 video resolution.

I think these pictures and videos show that the GH1 clearly has more resolution than the GF1. Now how noticeable that resolution difference is in a real world example(ie: Streaming on the net) is definitely debatable. However, when viewing the files straight out of the camera in their native format or upscaled to 1080p there is a clear and very apparent gain in the GH1 video pretty much across the board.

As of this moment I don’t see how you can beat the combination of a $650 GH1 body and the $400 20mm F1.7 lens for video. Shooting in manual focus mode and using the external mic jack yields the highest quality video I have seen.
 
lpowell said:
Thanks for posting these comparisons, mpgxsvcd. The GH1 clearly produces higher quality video frames than the GF1, and I think it's very likely due to a significantly improved sensor scanning algorithm in the GH1. But since the exact technique is apparently a trade secret that Panasonic hasn't yet disclosed, I'm not sure we can do much more than speculate at this point.
Click to expand...

I don't think knowing the technique they are using is as important as knowing what the results are. The GH13 is clearly better than the GF13 in every aspect. Let's just hope the GH2 is better than the GH13 and the GF13 for that matter.

If the GH2 is not significantly better than the GH13 then I will definitely be purchasing a GH1 to use along side my GF13. I would love to have two bodies to choose from to limit switching lenses. I switch lenses quite often right now.
 
mimirsan said:
Thats some interesting results!
My 2 cents...when I upscale the 720p in the pitiful editing programs I use it looks a bit crappy...so I use 1440x1080 setting in cam...saves me a lot of bother and looks great....a lot better than the 720p...so I guess at the end of it all...its down to what works best for the individual. :-)
Click to expand...



How do you set the GH13 to record 1440x1080?
 
mpg, can you update these opinions with your lpowell anamorphic results ? I believe you found the lpowel settings to give a substantially better result with the GF1.
 
You must log in or register to reply here.
Back
Top