Tech Note Why The Panasonic Had Poor Resolution

BobDiaz

BobDiaz

The Professor
Trying to read all the messages in this forum is almost a full time job. Every now and then, a message appears that stands out like a red flag. As most of you know, the Panasonic's resolution was measured at around 550 lines. See Adam Wilt's report:

http://www.dv.com/features/features_item.jhtml;jsessionid=GXFHKSPIX0ZJ0QSNDBCSKH0CJUMEKJVN?category=Archive&articleId=177103305&_requestid=392821&_requestid=93743

The key words to Adam's report are:
"The HVX's resolution numbers are suspect and should not be trusted until further testing is done."
Click to expand...

The big question is: WHY did the Panasonic do so poorly in the resolution test? I really wish I could have been there to see exactly what the settings of the camera were. Without a camera, all I can do is guess at a reason... Then I came across this message:

http://www.dvxuser.com/V3/showpost.php?p=418684&postcount=358

And then there is the stop. Checked my notes and the HVX was at a 5.6 during the resolution test. According to Barry, 4 or 2.8 would probably be better. Again, this is another "sweet spot" for the lens, aperture instead of focal length.
Click to expand...

The key line in the message jumped out like a red flag, "... the HVX was at a 5.6 during the resolution test..." The problem with F5.6 is that ANY 1/3" CCD HD camera is going to loose sharpness at this F Stop. IF you want to make that Super Sharp Canon look bad, just shoot a resolution chart at F5.6!!! Want to make it look worse than that, use F11.

SEE:
http://www.tvtechnology.com/features/Masked-Engineer/f_mario_orazio-08.17.05.shtml

The big news at NAB2005 was 1/3-inch HDTV cameras. That's what was at the front end of HDV camcorders from JVC and Sony as well as the P2 HDTV camcorder from Panasonic.

A 1/3-inch camera has a 6-millimeter image diagonal. In 16:9 HDTV, that's about a 2.9-millimeter image height. That means the Airy disk radius has got to be about 2.7 µm or less to get all the resolution of 1,080 lines.

So let's keep that 510 nm green wavelength and try f/5.6. So sorry! The Airy disk radius is about 3.5 µm, so the 1,080 lines will be diffraction limited. At f/4 you're okay on green, but red is still a small problem.
Click to expand...



IF the Panasonic had an F Stop of F5.6, the resolution was diffraction limited. Anyone who has an HVX can try this; shoot a resolution chart at F2.8, F4, F5.6, and F11. Once you pass F4, the resolution takes a dive.


Like I said, I can only guess at the reason, but this looks like a possible answer to the question: WHY did the Panasonic do so poorly in the resolution test?


I did say this in the message thread, "Discussion of dv.com results", but given the huge number of messages in that thread, it looked that the answer might get lost in a sea of messages.


Bob Diaz
 
Hey Bob, interesting point, not being up on the way digital imaging works to that degree, but I know in still cameras and film, if you stop you lens down, you get a sharper image, as you use less of the lens, and most lenses loose resolving power towards the edge of the element, thus the softening effect at image edges that can be found, especially on cheap lenses.
Or is it more of a case here, as you stop the lense down, its not the sharpness which is the issue, its the simple fact that the chip blocks 'film plane' is recieving less light, bent or otherwise, and therefore less TSA is exposed of the chips?? Thus reducing resolution???
 
Stopping down only works till we reach a point. That is true even in film cameras with SLR lenses but due to optics in this case. I notice that in 2/3 inch sensors there is only a slight loss at f12. 10% or so. I only have tried this in industrial cameras and cheap digicams though. Both with cheap lenses. I don't have a 2/3 inch camcorder.

I found something interesting here:

http://www.repaire.net/site/tournage/compar_Z1Sony_HD100JVC/comparatif_Z1_Sony_&_GYHD100_JVC.php

These cameras are also 1/3inch and sony has 1080pixels vertically.

The Sony does 800 vertical lines at f8:

http://www.repairevideo.net/imgrepaire/comparatifhdv/Mire04.jpg

Why would the HVX be diffraction limited at a wider f of 5.6? It's not logical. We don't really have to search for an excuse to justify the low resolution of this camera. Perhaps it is slightly better at f4. But it does not look like a diffraction limit. There is aliasing etc.
 
Basically stopping down would give a sharper image, but when the light has to pass thru too small a hole the rays start to bend, or the ratio of bent rays to the straight ones gets to be too big. Diffraction. In 35mm film (still) cameras the diffraction starts to lower the resolving power at around f 16 especially with the best lenses (bacause only good lenses are limited by diffraction...). There would be no point in using f64 with 35mm film, something that is normal with large view cameras. With smaller sensors the math starts to hit us even sooner like Bob here explanes. There is really no way to make a small sensor HD camera, unless you are ready to use it with only large apertures. Really there should be no smaller f-stops than 5.6 in the dials, then just a pack of ND filters...

And, there is nothing engineers can do about this, it is simple math: wavelength, focal length, aperture, sensor size.
 
I disagree, "for example" If you heard the HD100 was tested and verifed to be displaying 700 lines, then another test came through and read it was 500, would you not wonder why the differences. I would would want see more tests to understand the differences.

Well, I guess we'll all know soon enough.
Barry mentioned he would be doing a test with different settings and also
the effect of detail coring at different settings. It was believed to be set to +3.

We have seen recent results and have heard recent info that conflicts the HVX results done a couple weeks ago.
YES, I KNOW the testing was done by some VERY TALENTED INDIVIDUALS, but even themselves questioned why the HVX came in so low.

Steve
 
Differences will appear since it is possible to choose a shot so that alias appears as resolution. A video of a moving chart will show how aliasing appears.

From what I see in the chart, the HVX is not difraction limited at f5.6. Diffraction limiting appears as simple blur. If you see lines moving arround in the moving chart video or if you see a thin sharp line at the 800line point on a camera, what you see is alias. It is simple really. The trumpet starts with 5lines. If at any point you see 2 or 3 or 4 or lines breaking up, then you have alias, true ccd resolution limits. If the 5 lines go into blur at a specific point, then the system is diffraction limited and opening the apperture will help it resolve more and get higher MTF scores. A diffraction limited system is actually a good thing. A perfect lens is diffraction limited. One with abberations of all kinds is not.

 
quote quoted by Bob Diaz:

A 1/3-inch camera has a 6-millimeter image diagonal. In 16:9 HDTV, that's about a 2.9-millimeter image height. That means the Airy disk radius has got to be about 2.7 µm or less to get all the resolution of 1,080 lines.

It would appear that this quote and the logic that follows is based on the 1/3 inch CCD being pixel for pixel HD-- meaning that it has 1080 pixels vertically.

It appears to be well documented that these cameras all use pixel shift techology which would make the pixels roughly twice the 2.7 micron size implied here, would it not?

Bill Turner
 
Perhaps the reason panasonic is so secretive about their chips is that they opted for larger more sensitive pixels, with a lower count. ( so they still fit on the chip) maybe thats why the Dynamic range is good, and the color is good, but the sharpness is not there. They never did release a pixel count.

Funny anecdote.
I prepared a test with my aunt who knows nothign abot cameras, and showed her the frame grab jpegs, of the different cams, and she said the "Cine alta" looked the clearest, but the "HVx" had the prettiest colors.

said like a true professional.

I think your Theory is sound, cause this was a major point when these HD cams were first being introduced. The diffraction with small lenses is a problem. I mean a 1/3 camera is using a tiny lens in comparison to a SLR or even a 2/3 camera.
 
Right now, all I can work from is theory. A simple test would tell if the theory is correct or not. Just shoot a resolution chart at F2.8, F4, F5.6, and F8. The theory says that at F2.8 should be the sharpest image, F4 will be a bit less, but close, and F5.6 and F8 should be worse. Even if this test is different from the 6 camera test, the readings we are looking at are relative to the other F Stops and would show the impact of diffreaction.

Now if anyone has an HVX-200, 1080 monitor, lights, a resolution chart, free time, and lives within a reasonable drive to the L.A., CA area, I'd even come over and help you test it.


Bob Diaz
 
If i qualified in anyof thoes categories i wouldnt be here on DVxuser, id be shooting 1080P videos of my hamster on a chart, on hollywood boulevard.........
 
I've tried this and there is very little diffference between 2.8, 4, 5.6 and 8. Past 8 you can see a difference but the other apertures seem to work equally well. My testing so far puts the HVX right in the 550-600 LPH range, maybe a hair more than Adam recorded. Don't fret though, this is still substantially better than SD and the images look very good and filmish on a good HD monitor. I put my HVX against an XL-H1 at Samy's the other day. Both on 50" HD plasmas, the XL was a hair sharper but much dirtier, noisier and less pleassing in my opinion. And I did play with the XL's settings, low knee coring, all of that, it was still noisy, even in good light. HVX is still the better camera.
 
I've tried this and there is very little diffference between 2.8, 4, 5.6 and 8. Past 8 you can see a difference but the other apertures seem to work equally well.
Click to expand...

OK up to F8??? Boy, that IS a real surprise. :shocked:

That would put the effective pixel resolution at around 1024 x 560. However, before anyone runs out and says the pixel resolution is ... a lot more tests would need to be done.

It would be interesting to take the Canon and measure it at F2.8, F4, F5.6, F8, and F11. Then plot F Stop vs. resolution.

Next compare it to a plot done on the the JVC.

Last of all a plot done on the Panasonic running 1080 and another running 720.

Too many ideas, not enough time or equipment, oh well.
In the end, so what, if the image looks great, who cares...



Bob Diaz
 
You can't test the XLH1 at 11, because it doesn't go to 11. Its iris is limited to 9.6.

The JVC has that capability. The Sony goes to 11 as well.
 
Evinsky,
Thanks for checking the various f-stops.
Did you happen to play with the detail coring and check its effect on the rez chart?

Hmm, kind of good news and bad news, atleast up through F8 we should not have to be worried about diffraction limiting. :)

thxs, Steve
 
Barry_Green said:
You can't test the XLH1 at 11, because it doesn't go to 11. Its iris is limited to 9.6.

The JVC has that capability. The Sony goes to 11 as well.
Click to expand...



Sounds like a Nigel Tufnel quote.
 
In a perfect lens, optical theory (governed by the laws of physics) says that maximum resolution will occur at maximum aperture (wide open). In the real world aberrations (the lens is NOT perfect) prevent this. Many aberrations are reduced as a lens is stopped down (some such as chromatic are not).

As a real "imperfect" lens is stopped down the resolution increases until the degredation caused by diffraction becomes the limiting factor and then it begins to decrease.

A key factor in the perception of all this is the limiting resolution of the media receiving the image. If the lens resolution exceeds that of the sensor no degredation will be seen until the resolution falls below that of the sensor.

Another thing to know is that the diffraction degradation due to the scattering of the light by the edges of the aperture depends on the actual physical size of the aperture being relatively small.

In a simple lens there is a very direct relationship, for a given f number (4 for example) as the focal length gets shorter the opening gets smaller. The effects of diffraction will show up much sooner in a short focal length lens than a long telephoto (f 32 might be a very useable stop in a very long telephoto while stopping a very short focal length lens down past f8 might cause a significant drop in resolution)

A modern zoom lens is a much more complex design, clearly the actual size of the aperture at f4 does not change as the lens zooms from long to short, but the apparent size (looking at it from the front of the lens thru all the elements in front of it) does.

At the end of the day it is the picture that counts, but clearly with these small sensors avoiding stopping down too much should produce better results all things being equal.

Bill Turner
Scheider Optics
Century Division
 
Barry_Green said:
You can't test the XLH1 at 11, because it doesn't go to 11...

...The Sony goes to 11...
Click to expand...
jimagine said:
Sounds like a Nigel Tufnel quote.
Click to expand...


I can see their new ad campaign now:


sony.jpg
 
Bob,

If lower F-stops = greater resolution, would it be better to stop the camera down using the ND filters and open up the iris on a sunny day?

Jes
 
You must log in or register to reply here.
Back
Top