AG-DVX200 4K Large-Sensor Handheld Camcorder

This camera and the recent footage shows very little of actual detail:

View attachment 103622
Brrr.. I have Blackmagic 4K and as soon as possible will get URSA Mini 4.6K for my Cinema production, but I'm also would like to get good ENG-type of camera for something simple to shoot and was thinking about Canon C-100 mark II because I like the image and auto focus in it ... but when I find about this Panasonic camcorder I like it's options, and trying to find out what is it, and when I download and watch this footage I see it's looks terrible...:thumbdown no details, looks like a little garbage all over the screen, Bokeh is nothing at all... and in other footages, without grading Color and Dynamic Range looks bad...
let me try to explaine what I mean:
What I like in BMPC4K images is a details, but some how they are smooth !!! not so harsh as details in Sony FS700-FS7 cameras... ->

BMPC4K-RAWs.jpg


Original - http://www.sergecinema.com/temp/BMPC4K-RAW.jpg

Frame_4K_1s.jpg


Original - http://www.sergecinema.com/temp/Frame_4K_1.jpg

I don't know if my limited English are able to explain what I mean.. When I compare this footage with even 1080p image from Canon C-100 Mark II camera I feel that Canon image is much better!...

What a reason to make so Ergonomicaly GOOD 4K camera body with so poor image quality? ..so, I don't really understand what this camera made for? :cry:
 
Brrr.. I have Blackmagic 4K and as soon as possible will get URSA Mini 4.6K for my Cinema production, but I'm also would like to get good ENG-type of camera for something simple to shoot and was thinking about Canon C-100 mark II because I like the image and auto focus in it ... but when I find about this Panasonic camcorder I like it's options, and trying to find out what is it, and when I download and watch this footage I see it's looks terrible...:thumbdown no details, looks like a little garbage all over the screen, Bokeh is nothing at all... and in other footages, without grading Color and Dynamic Range looks bad...
let me try to explaine what I mean:
What I like in BMPC4K images is a details, but some how they are smooth !!! not so harsh as details in Sony FS700-FS7 cameras... -L

I don't know if my limited English are able to explain what I mean.. When I compare this footage with even 1080p image from Canon C-100 Mark II camera I feel that Canon image is much better!...

What a reason to make so Ergonomicaly GOOD 4K camera body with so poor image quality? ..so, I don't really understand what this camera made for? :cry:

Noel Evans who has shot this camera says that the detail is excellent. Look earlier in this thread. His reputation is superb and I value his opinion.
 
This camera and the recent footage shows very little of actual detail:

View attachment 103622


Have to address a few points. One above for starters - what was this taken from?

I chose not to put out any images I shot because I felt there was a lens alignment issue that was causing the left side of the image to soften. I've since confirmed this was actually the case.

But I can confirm the level of sharpness I was able to get out of this camera was as good as any I have seen - without looking like a faux sharpened image. Case in point (FWIW Im only sharing this image to make that point). Shot is focused on the stadium from which Im standing a good 3-400 metres from. Lots of very fine lines - its retained nice straight edges that look natural. Ignore the left side with the big red 7 - thats the lens on that side.

EDIT - BAH HUMBUG. DVXuser image upload is restricted - forgot about that. I will upload a 4k image later to my website and relink here.

ETIHAD 2.jpg

I tested the camera on an interview style set up as well - heres what I ran for that. Audio direct to cam was very good. I didn't record anything to shogun.

IMG_9906.jpg
 
Hi, I've seen this referred to before and I know to some extent how color subsampling works (I'm an engineer working in broadcast). I don't get the 8b 420 -> 10b 444 path, do you care to reference the processing path and which hw/sw does it reliably. thanks.
Any proper NLE software should do it properly.

In 4:2:0, there's one color sample for each 2x2 block of pixels. So in a 3840x2160 frame, the color matrix is 1920 x 1080. If you downscale the luminance from 3840 x 2160 down to 1920 x 1080, then you end up with a luminance matrix of 1920 x 1080 and a color matrix of 1920x1080. You have one color sample for every brightness pixel. Hence 4:4:4.

With 8-bit pixels, you get 256 levels of brightness; with 10-bit pixels you get 1,024 levels of brightness. If you take four 8-bit pixels and sum them together, you get one "big" pixel that has 1,024 levels of brightness. So when you downconvert from 3840 x 2160 down to 1920 x 1080, you sum the pixel values and end up with 1,024 distinct levels of brightness in each pixel, which is what 10-bit delivers.
 
This camera and the recent footage shows very little of actual detail:
The camera shows astonishing levels of detail. I cannot speak to whatever post process that footage went through, but I can say that the DVX200 is extremely sharp. In the past Panasonic may have made cameras that weren't quite as sharp as some of the competition, but that is absolutely not the case with this camera.

This is just one example; here's a 4K res chart from a Red One, and a 4K res chart from the DVX200. Pixel-for-pixel extracts, shooting the exact same chart. The Red One was using a $13,000 Ultra Prime 50mm lens.

Red One res extract.jpgDVX200 Res extract.jpg
 
Last edited:
Sony RX10/RX10 II 24-200 at F2.8. Fixed lens camera with constant aperture.
And a smaller chip, and its zoom range is only 8:1 instead of the DVX200's 13:1 zoom. Isn't the same.

Furthermore, has anyone tested it to see if it actually truly is a fixed aperture? Sony has fudged that in the past; they marked the EX1 as "F/1.9", which led many people to say it was a fixed aperture throughout the zoom range, but it absolutely was not, it ramped down just like every other camera's lens did. I'm not saying they're misleading anyone here, only that I wouldn't be so quick to say that it is definitely a constant F/2.8 without actually testing it to see. The EX1 certainly wasn't.
 
What a reason to make so Ergonomicaly GOOD 4K camera body with so poor image quality?
They didn't. They made an ergonomically good 4K camera with excellent picture quality. Then, unfortunately, they put it in the hands of people who didn't know how to use it, who had no experience with it, and those people shot whatever they shot, and then did whatever they did in post, and the results are underwhelming. No surprise there. I'm sure they are all very talented shooters, but the DVX200 is an incredibly complex camera, far more advanced than any Panasonic handheld that's ever been released before (with the possible exception of the PX270). I would not expect anyone to pick this camera up for the first time and be able to produce the best results from it.

Give those same people a few weeks with the camera, and they'll probably turn out vastly better results. As I've said repeatedly, no purpose is served by looking at this preproduction footage from an engineering sample, and I think it was a mistake for this footage to even be released. They should have embargoed the footage until they had a finished camera that properly represents what the final product will look like, and they should have given the shooters long enough to get familiar with it and learn its ins and outs before expecting proper representative footage from them.
 
Does this apply to throwing UHD footage into a 1920x1080 timeline? Or exporting a UHD timeline/project and compressing it to 1920x1080? Or both?

Any proper NLE software should do it properly.

In 4:2:0, there's one color sample for each 2x2 block of pixels. So in a 3840x2160 frame, the color matrix is 1920 x 1080. If you downscale the luminance from 3840 x 2160 down to 1920 x 1080, then you end up with a luminance matrix of 1920 x 1080 and a color matrix of 1920x1080. You have one color sample for every brightness pixel. Hence 4:4:4.

With 8-bit pixels, you get 256 levels of brightness; with 10-bit pixels you get 1,024 levels of brightness. If you take four 8-bit pixels and sum them together, you get one "big" pixel that has 1,024 levels of brightness. So when you downconvert from 3840 x 2160 down to 1920 x 1080, you sum the pixel values and end up with 1,024 distinct levels of brightness in each pixel, which is what 10-bit delivers.
 
barry, you seem like just the awesomest guy ever. wanna get a drink? i dont drink so ill have cocoa while you get plastered. also youll have to fly to houston.
 
Brrr.. I have Blackmagic 4K and as soon as possible will get URSA Mini 4.6K for my Cinema production, but I'm also would like to get good ENG-type of camera for something simple to shoot and was thinking about Canon C-100 mark II because I like the image and auto focus in it ... but when I find about this Panasonic camcorder I like it's options, and trying to find out what is it, and when I download and watch this footage I see it's looks terrible...
D:\TMP\msohtmlclip1\01\clip_image002.png
no details, looks like a little garbage all over the screen, Bokeh is nothing at all... and in other footages, without grading Color and Dynamic Range looks bad...
let me try to explaine what I mean:
What I like in BMPC4K images is a details, but some how they are smooth !!! not so harsh as details in Sony FS700-FS7 cameras... ->
…
I don't know if my limited English are able to explain what I mean.. When I compare this footage with even 1080p image from Canon C-100 Mark II camera I feel that Canon image is much better!...


Your English is good enough. If some people don't get what you mean is because of they look differently at movies – not because of your English.

You are definitely not alone in preferring the output of the Canon C cameras over the Panasonic and Sony ones. Not all people see the difference between harsh details caused by over-sharpening and organic looking abundance of details. Not all people note the differences in the color rendition between those cameras – and when they see the differences, they prefer the Panasonic and Sony colors over the Canon ones.

What a reason to make so Ergonomicaly GOOD 4K camera body with so poor image quality? ..so, I don't really understand what this camera made for?
D:\TMP\msohtmlclip1\01\clip_image003.gif


Possibly because of:

  1. So many people do like the images from their cameras (though definitely not all people).
  2. They cannot do any better at the designated price tag. Concerning color science, they cannot do any better at any price.
  3. They look at the images like so many of their clients do – not seeing the difference between harsh details caused by over-sharpening and organic looking abundance of details.
 
…
With 8-bit pixels, you get 256 levels of brightness; with 10-bit pixels you get 1,024 levels of brightness. If you take four 8-bit pixels and sum them together, you get one "big" pixel that has 1,024 levels of brightness.
…


1,024 levels of brightness which have only 256 *different* levels of brightness. You cannot 'create' brightness differences that weren't recorded by the camera. One would get the very same effect putting 1080p 8 bit footage on 10 bit timeline – which brings no practical benefit.
 
1,024 levels of brightness which have only 256 *different* levels of brightness. You cannot 'create' brightness differences that weren't recorded by the camera. One would get the very same effect putting 1080p 8 bit footage on 10 bit timeline – which brings no practical benefit.

Wrong! If you downrez UHD to 1080, 4 pixels of the UHD picture (2x2 block) will combine to make 1 pixel of the 1080 picture. So at the dark end, if the 4 UHD pixels are 0,0,0,0, then the 1080 pixel value is 0. If the 4 UHD pixels are 0,0,0,1, then the 1080 pixel value is 1. And so on until you get to the brightest pixel at 255,255,255,255 which combine to 1020. So basically, 8-bit UHD data can be downrezzed to 10-bit 1080 data, with 1021 unique levels of brightness.
 
Back
Top