Fantastic Panasonic GH2 video

Kholi said:
Pappas! I've been good. My ghastly attendance is due to the feature film, but I've got a few days "off" so to speak so I'm just catching up on all the geek-speak.

I probably won't be purchasing an AF100, I'd like to use one soon, however!
Click to expand...

Kholi! Good to know your doing well. Creative projects, especially larger ones will suck your time up quick that's for sure, however that's a good thing compared
to not being on a project.

Me neither- the AF100 is a cool camera that I'm looking forward to trying out. However I don't need to own one.

Pappas
 
mrgroove said:
hello, i'm Emmanuel Pampuri ...
Click to expand...

Greetings, Emmanuel. I don't have a question right now, but I just wanted to properly thank you, Matthieu Misiraca, and the entire crew (lighting, makeup, model, etc.) for the wonderful work you all did -- but especially for your sharing it with us. Bravo & cheers!
 
mrgroove said:
If you have some questions, my english is not very good but i can try to gives you some answers.
Click to expand...

Thanks for joining the discussion. Great video! Looks amazing! With the Nostalgic setting did you stay with the default, or did you adjust the contrast, sharpness, saturation, and NR at all? Also, did you white balance or just set the temperature with the K settings? Thanks!
 
oh my gosh, this is a great example. It helps out my theory that instead of skipping pixels while reading the sensor, it's capturing the whole sensor in video mode and downscaling it to hd using a bicubic or similar higher quality scaling algorithm. Even though the sensors SNR ratio isn't as good as the ex1 I currently own and other similar cameras, it's downsampling the image from such a larger sensor resolution that like when down-scaling a picture, the lighter noise all but disappears and you gain so much sharpness.

can anyone confirm/deny the gh2 doesn't just skip pixels to get an hd resolution out of the sensor, but its capturing each pixel and downscaling? if true it might persuade me to pick one up.
 
It's always been a mystery on what these DSLRs are doing to their image when recording but whatever the GH-2 is doing....it's doing it much better than any other DSLR...that's for sure.
 
Fohdeesha said:
can anyone confirm/deny the gh2 doesn't just skip pixels to get an hd resolution out of the sensor, but its capturing each pixel and downscaling? if true it might persuade me to pick one up.
Click to expand...
Depends on if he was in ETC mode or not.

In ETC mode it appears that it just samples a 1920x1080 patch at full res out of the center of the 18-megapixel chip.

In full mode, it does the same thing the GH1 did, because it has the same level of luma aliasing tha the GH1 had.
 
Fohdeesha said:
oh my gosh, this is a great example. It helps out my theory that instead of skipping pixels while reading the sensor, it's capturing the whole sensor in video mode and downscaling it to hd using a bicubic or similar higher quality scaling algorithm. Even though the sensors SNR ratio isn't as good as the ex1 I currently own and other similar cameras, it's downsampling the image from such a larger sensor resolution that like when down-scaling a picture, the lighter noise all but disappears and you gain so much sharpness.

can anyone confirm/deny the gh2 doesn't just skip pixels to get an hd resolution out of the sensor, but its capturing each pixel and downscaling? if true it might persuade me to pick one up.
Click to expand...

I think a big hint about how the GH2 samples it images is the 4 megapixel 40 fps photo mode. I think it is how the gh2 samples one quarter of pixel, that is one every four ones for the de bayer and then downscale it to 1080p. While the other Dslr just debayer at a one to one level. That is why you have much lower aliasing and moire. If it was sampling all the pixel of the sensor, I think we would have no aliasing and moire as the olpf is tune to that resolution. Further proof of that is the tele version, which uses the center window of the sensor, so no aliasing and moire as it is a one to one sampling, but a little soft because of bayer pattern you get less resolution at the pixel level.
 
mrgroove said:
hello
i'm Emmanuel Pampuri,
i can see here that some of you likes my little video ... i'm so happy thx a lot !
of course this video is nice because i used the buetiful light made by matthieu misiraca for the fake ad we shot with ARRI Alexa and RED One MX.
I Confirm the nostalgic mode and no grading.
This GH2 is the most incredible "Photocam" i had ever used !
If you have some questions, my english is not very good but i can try to gives you some answers.
Click to expand...


Merci pour la vidéo, c’était un cadeau de noël et je n’ai pas pu m’empêcher de le partager ici. C’est vraiment un super travail et le premier exemple montrant vraiment les qualités du gh2 en reportage mais surtout pour moi sur un plateau avec lumière etc. J’attends avec impatience les conclusion sur la dynamique et sans prétention les images du Red et de Alexa.
 
Barry_Green said:
In ETC mode it appears that it just samples a 1920x1080 patch at full res out of the center of the 18-megapixel chip.
Click to expand...

Barry, shouldn't we expect then color resolution to drop in ETC mode compared to full res mode? This since in ETC mode there will be no oversampling of photosites relative to output pixels, just a one-to-one correspondence. Assuming a Bayer color filter pattern, having 1920x1080 photosites would not be enough to get reasonable color resolution for HD output.

I don't know the sensor size but maybe the center patch in ETC mode (I wonder if it roughly corresponds to the image a c-mount lens can project) contains more than 1920x1080 photosites (2 mega), maybe 3 or 4 mega photosites, a number big enough to get a reasonable good interpolation of color, and small enough to avoid severe line skipping and pixel binning.
 
lexicon said:
Barry, shouldn't we expect then color resolution to drop in ETC mode compared to full res mode? This since in ETC mode there will be no oversampling of photosites relative to output pixels, just a one-to-one correspondence. Assuming a Bayer color filter pattern, having 1920x1080 photosites would not be enough to get reasonable color resolution for HD output.
Click to expand...

Bayer pattern is enought for 4:2:2 :-)
 
Merci Daniel pas de souci si mes vidéo sont disponible sur vimeo avec les sources téléchargeables c'est pour que la comuneauté en profite et que l'on puisse partager, échanger ...

Transtation ( i hope you can understant it ;-) )
Thank you Daniel, i'm ok to share my video, that's why i choose Vimeo, i hope my work will be shared a lot and all the video / filmaker comunity enjoy it ...
 
mrgroove said:
Hi Mike, i didn't change any setting just using the "Nostalgic" preset. I shot with K temperature seting
Click to expand...

So you did not reduce the contrast setting and so we can conclude that we could get even more dynamic range. Even if it was not in a very contrasty condition, the dynamic range seemed quite good. I am really getting a bad feeling about the Af-100, I am seeing highlights clipping in very low contrast scene!!!
 
mrgroove said:
Thank you Daniel, i'm ok to share my video, that's why i choose Vimeo, i hope my work will be shared a lot and all the video / filmaker comunity enjoy it ...
Click to expand...

What a wonderful attitude. Thank you so much and yes, I think many people are enjoying your work. I know I am.
 
lexicon said:
Barry, shouldn't we expect then color resolution to drop in ETC mode compared to full res mode? This since in ETC mode there will be no oversampling of photosites relative to output pixels, just a one-to-one correspondence. Assuming a Bayer color filter pattern, having 1920x1080 photosites would not be enough to get reasonable color resolution for HD output.
Click to expand...
Not necessarily. Bayer can deliver a reasonable 4:2:0 color sampling at the native resolution, since 4:2:0 means that the color sampling is one color sample for each 2x2 block of pixels, and that's pretty much exactly what a Bayer 2x2 block is.

In a 1920x1080 video, the chroma in 4:2:0 is going to be recorded as 960 x 540. In a 1920x1080 Bayer patch, you'll have 960x1080 green, 960x540 blue, and 960x540 red. So it's enough.

What you're not going to get is a full 1920x1080 resolution, because the demosaic process is going to throw some of that native resolution away.

I don't know the sensor size but maybe the center patch in ETC mode (I wonder if it roughly corresponds to the image a c-mount lens can project) contains more than 1920x1080 photosites (2 mega), maybe 3 or 4 mega photosites, a number big enough to get a reasonable good interpolation of color, and small enough to avoid severe line skipping and pixel binning.
Click to expand...
Well, we know that the entire chip is 4900 pixels. And that the "effective crop" is 2.6. And 4900 / 1920 = 2.55. Rounded up, that's 2.6. So I'm comfortable in saying that it's a 1920x1080 patch out of the center of the chip. Seems reasonable.
 
Danielvilliers said:
So you did not reduce the contrast setting and so we can conclude that we could get even more dynamic range. Even if it was not in a very contrasty condition, the dynamic range seemed quite good. I am really getting a bad feeling about the Af-100, I am seeing highlights clipping in very low contrast scene!!!
Click to expand...

I thought the same thing at first, but upon closer examination, it seems to be exhibiting the same lower DR than expected as the rest of the videos shot with this camera. It seemed to be capturing plenty at first, then I realized they were outside shooting on an overcast day and a lot of things are still on the verge of blowing out. On some of the indoor shots things are really clipping prematurely it seems. I really wish someone would get this thing tested for DR in video mode, because in stills mode it's a whopping 11.5 stops. But for some reason I'm just not seeing that in video mode :(
 
Fohdeesha said:
I thought the same thing at first, but upon closer examination, it seems to be exhibiting the same lower DR than expected as the rest of the videos shot with this camera. It seemed to be capturing plenty at first, then I realized they were outside shooting on an overcast day and a lot of things are still on the verge of blowing out. On some of the indoor shots things are really clipping prematurely it seems. I really wish someone would get this thing tested for DR in video mode, because in stills mode it's a whopping 11.5 stops. But for some reason I'm just not seeing that in video mode :(
Click to expand...

I think some of the shot are a bit overexposed by the shooter because it is a new camera and run & gun. He did not touch, grade the footage in anyway. He could surely shoot a little darker to get less clipping and the roll-offs are still quite nice. Footage with the sun are quite rare until now. It seems that Until now shooting people outside is not in vogue, the 3200 iso are all the rage.
 
You must log in or register to reply here.
Back
Top