Tech Note Why The Panasonic Had Poor Resolution

ullanta said:
Also, given the Codec issue, we can assume that this is a RECORDED format chart (i.e., compressed); do we know if the ones posted by real-stream were the same?
Click to expand...
The one posted by reel-stream is in fact this very exact same recording! The difference was in the post process to get it extracted and displayed, but both of them came from the same MXF file.
 
Barry_Green said:
The one posted by reel-stream is in fact this very exact same recording! The difference was in the post process to get it extracted and displayed, but both of them came from the same MXF file.
Click to expand...


Ooops, I mis-spoke (er, mis-posted) - I meant the charts shot by other cameras on the real-stream site, not the HVX one!
 
Danilo Del Tufo said:
Considering this chart, what is the resolution of Hvx200?I dunno how reading the chart. Thanks!
Click to expand...

I read it as 600-650 on the horizontal, 700-800 on the vertical (depending on generosity/standards).

Also, I noticed on Kaku's funky res charts that the HVX seemed to have better resolution on the diagonal trumpets than on either the horizontal or vertical. The diagonals were extremely clean! I haven't seen charts from other cameras with diagonals... but I wonder how all the tradeoffs with pixel-shift, etc, work there. Some of the perceived vs. "chart" resolution issues the HVX has might be due to the fact that it handles a different set of angles better? Or is this laughable?
 
We posted the exact website we got the rest of the charts from. It's a pretty well-known site, their tests on the Z1 and HD100 were relatively thorough as they tested different iris apertures and their effects on resolution. Although their resolution readings are a bit off in my opinion, i just posted the charts, not what I think the numbers are.

The Andromeda DVX100A frame is certainly the image you record.

As far as the other cameras(HD100 and Z1), are you asking if the images where taken off the HDV tape, or off the analog out?

Cheers,
Juan
 
pertierr said:
As far as the other cameras(HD100 and Z1), are you asking if the images where taken off the HDV tape, or off the analog out?
Click to expand...

Yeah, that's what I'm wondering....
 
I dont really get into spec testing, not my thing but some guys who are very into it at one of the production companies I work with have been shooting charts and they have not gotten the rez more than 600, usually hits at around 550-580.. they are reading it more like Adam Wilt, and not stretching that number a bit. We are all confused because in motion it looks like more rez than that. Something weird is going on inside... still looks great though...



ash =o)
 
I may be wrong, but the footage looks consistent with what i've seen off standard HDV recordings from these cameras. It doesn't make a note anywhere in the article that the testers went through the trouble of recording the analog out, so I'd assume it's off the HDV tape on both the HD100 and Z1.

Cheers,
Juan
 
limo991 said:
I see the "Jan statement" thread is locked and it had 1000 views. This is definetely a hot subject. Imagine if Jan actually responded and released some info!
Click to expand...
I was saddened to see it shut down. I was asking a question which had completely factual statments in it, and I feel this needed a factual response, not a "CLOSED" stamp on it.

To be honest I'm afraid to ask again, very touchy subject indeed! :(
 
Yes Bob, we are just speculating here. Don't take this the wrong way:) Pixel shifting generally gives x1.3 to x1.35 on the final output. 1.5 was a speculation on a maximum. In experiments I have done with models, 1.35 produces a nice artifact free and smooth result if we want a trully videophile (is there such a term?) result.
 
On the static chart Barry posted I see a nice usable sharpness setting and about 650 horizontal and 700 vertical. I can't be sure though where aliasing starts on a single static chart. From my personal tests it starts where curviness appears in lines. This is close to what 1080 HDV cameras do in interlaced mode in static test charts. And much better than what they do in the vertical resolution progressive modes.

I find it strange how aliasing on vertical axis can change that much from one shot to the other. It's not something that should change with focal lenght or apperture.

I hear a lot about the horizontal performance and how it should be higher. On a 1280x1080 progressive format, the horizontal performance on the chart should be 1280/(1080*16/9)=0.67 or 2/3 of the vertical performance. The theoretical maximum is 720horizontal and 1080 vertical lines.

If all companies used the same implementation, and a 1440x1080 camera did 800lines horizontal and vertical, the 1280x1080 camera would do about 710 horizontal and 800 vertical.

The PAL HVX 1080p format will be 1440x1080? Anyone know about that? Barry? If that is true, it should be close to the Canon in performance and slightly better than the Sony. Hope it does not have the Canon mosquito-like noise:)
 
Mr. Blonde said:
I see Barry's chart reading 600H x 580V.
Click to expand...

I think you must be switching H and V... there's a lot of wiggle room for the actual numbers, but I think the vertical resolution (horizontal "trumpets") is clearly much higher than the horizontal...?
 
limo991 said:
Yes Bob, we are just speculating here. Don't take this the wrong way:) Pixel shifting generally gives x1.3 to x1.35 on the final output. 1.5 was a speculation on a maximum. In experiments I have done with models, 1.35 produces a nice artifact free and smooth result if we want a trully videophile (is there such a term?) result.
Click to expand...


I'm finding that around 1.3 to 1.4 is the ideal increase for pixel shift, but one can push things to the max and use 1.6 as the factor. With 1.6, the results are less then ideal. So, for fun, let's assume 1.3 to 1.4 for the range.

The Kell Factor is another problem I haven't worked out yet. One can argue that 0.6 gives reasonably clean results, but 0.7 could be used too. I know that when pushed, 0.85 is used, but things are getting a little fuzzy with this number.

Taking both pixel shift and Kell into account, the range is something like this:

1.3 x 0.6 = 0.78 low end (close to 0.8)
1.4 x 0.7 = 0.98 high end (close to 1.0)

Say the mid point is 0.9
​
If we look at a resolution chart and take the reading H or V and multiply by 0.9, that should get us in the general range of the native resolution of the CCD. +/- a million percent :cheesy:

Great, but given that fact that then numbers are from 540 to 700, which is the correct number to use? This all come back to the three great questions in life:

  • (1) What is the meaning of life?
  • (2) Is there really a God?
  • (3) What is the native resolution of the HVX-200 CCD's?
At this point, Jan is reading this thinking, "You guys have NO life!!!" :happy:


Bob Diaz
 
ullanta said:
I think you must be switching H and V... there's a lot of wiggle room for the actual numbers, but I think the vertical resolution (horizontal "trumpets") is clearly much higher than the horizontal...?
Click to expand...
ullanta, with horizontal being left to right and vertical, up and down. I clearly see the horizontal being slightly higher than the vertical. :shocked:
 
Mr. Blonde said:
ullanta, with horizontal being left to right and vertical, up and down. I clearly see the horizontal being slightly higher than the vertical. :shocked:
Click to expand...

Ummm.... the horizontal lines give you vertical resolution; and the vertical lines give you horizontal resolution.

Horizontal res = the number of vertical lines you can distingush
Vertical res = the number of horizontal lines you can distinguish.

Think of it this way... with only horizontal lines extending all the way across the screen, you'd have NO info about horizontal resolution at all; but how many of these you could put on the screen and still see would tell you a lot about vertical res.

But then we're saying the same thing - "the horizontal lines/trumpets indicate more resolution than the vertical lines/trumpets". Which is cool.
 
It's the diagonals that intrigue me... but noone else seemed to have any comment... and most charts don't seem to have diagonals...
 
From what I recall, Kaku's charts showed a higher diagonal than vertical or horizontal.

Is that what u mean ullanta?
 
At the same focal length the HVX image is noticeably wider than the DVX. Which makes sense since the 16x9 chip is wider than the 4x3 chip.
 
You must log in or register to reply here.
Back
Top