Discussion of dv.com results

I think the HVX200 does not look bad. The image is very nice just lower resolution. In the tests, the lens at appears to have less CA than the other lens but the camera appears to be EXTREMELY sensitive to overexposure. In the frame with the girl, the cameras have similar exposure and some show CA (Canon and Cinealta mostly) and my JVC has lots of CA in other settings, but the HVX has got this annoying color fringing flowing from overexposed areas. It is worse than some very cheap camcorders in that department. This will probably keep the user from overxposing on purpose to get better tonality in mid tones on this 8bit format:(

Can you shoot a tree or building and intentionally overexpose the sky Barry? This is a typical test for this problem.

Wide angle and wide open and small imaging areas make this problem worse.
 
F5.6 is Diffraction Limited

F5.6 is Diffraction Limited

Jay Nemeth said:
Absolutely. It's impossible for a zoom lens to focus the three primary colors at the same distance throughout the entire zoom range. Usually a point somewhere in the middle towards wide end is picked by the lens manufacturer, and this is what DPs call the "sweet spot". At that focal length, the red, blue, and green portions of the image are hitting exactly where they should after going through the prism and to the imager. As you zoom out or in, chromatic abberations start to show up as the glass just isn't able to keep all the colors tightly wound up at all focal lengths.

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.

As far as the chart setup, I spent a great deal of time making sure all the cameras were at the same height, the chart was the same height as the lenses, the chart was perpindicular top to bottom, and we re-aimed the chart to each individual camera so there would be no keystoning. All the chart framings were checked in underscan to make sure the carrots just reached the edge of the raster. Of course there was the wiggle test that Adam was doing, so they did get off from time to time, BUT the size of the chart relative to the field of view was not affected by this. I also made sure the area behind the cameras was dark so there would be no reflections or flare in the chart.

Had to setup a Gyron tonight, so I didn't get a chance to reshoot with the HVX. I'll post as soon as I do.

Jay
Click to expand...


FIRST READ THIS LINK:

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


IF the Panasonic or any other camera was set to F5.6 or higher, the resolution is going to be less than if the camera was set to F2.8!!!!!



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...




Anyone who has the Canon, try shooting a resolution chart at F2.8 and again at F5.6, you will see a BIG drop in the resolution.

NOTE TO BARRY: I know you are busy, but if you shoot a resolution chart at F2.8 and again at F5.6, the F5.6 should show a major drop in resolution. If you can shoot this and post JPEGs, that would help to show the diffraction issue.


If anyone must shoot in bright light, use the 1/8 or 1/64 Neutral-density filters built into the camera.

Bob Diaz
 
pertierr said:
As far as achieving a filmic look, I would say "latitude is everything".
Click to expand...

That's the thing that still has me thinking about the HD100. It had wider latitude on the skin tones in the test. And from everything I've read on the CA so far it appears to happen when the lens is at full telephoto and iris is wide open. That can easily be worked around.

I'm also NOT a fan of chroma noise like the HVX has. it's ugly to me. I don't want colors dancing in the noise on the shadow side of someone's face in a closeup... but maybe that can be overcome by shooting flatter and increasing contrast in post.

Not sure how wide angle the JVC lens is vs. the HVX - I think the HVX is pretty wide, which is good. Does anyone know how these compare?

So many variables. I think talented people will make any of these cameras look good, but I did hope the HVX would clearly have the best image quality and nothing I've seen indicates that's the case at all. They all look pretty good.

I also hoped I'd have an HVX in November.
 
yes bob.. good point that some of us realized and me and Barry have talked about it so i know he knows about shooting open. This has always been true, but never affected SD optics that much.

The HVX is really sensitive folks.. i was outside shooting in the open today with it and even at 1/64 i could used some more kick down.. I would definately add some ND's to everyones shopping lists.
 
BobDiaz said:
FIRST READ THIS LINK:

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


IF the Panasonic or any other camera was set to F5.6 or higher, the resolution is going to be less than if the camera was set to F2.8!!!!!

Anyone who has the Canon, try shooting a resolution chart at F2.8 and again at F5.6, you will see a BIG drop in the resolution.

NOTE TO BARRY: I know you are busy, but if you shoot a resolution chart at F2.8 and again at F5.6, the F5.6 should show a major drop in resolution. If you can shoot this and post JPEGs, that would help to show the diffraction issue.


If anyone must shoot in bright light, use the 1/8 or 1/64 Neutral-density filters built into the camera.

Bob Diaz
Click to expand...

Holy Cow Bob! How do you find this stuff? Excellent information to factor into getting the most out of the HVX or any 1/3" HD cam.
 
limo991 said:
but the camera appears to be EXTREMELY sensitive to overexposure.
Click to expand...
Only in the CineGamma modes. This is a known issue from back to the original DVX100 -- CineGamma has no "knee" circuit, so you always want to protect against blowouts.

If you're shooting for maximum dynamic range and cannot protect against the blowouts, you may be better off shooting "high" gamma with the knee circuit set on "low".

If you can protect against blowouts, cine-like-d is the better choice.

I wanted to use the knee on that shot, but I was asked to provide cine-like-d, so that's what we set it to. And the result is blue fringing around blown-out segments. You don't have to get that, you can shoot with different settings to prevent that, but that wasn't the way that test shot was requested.
 
PappasArts said:
This is also the same situation with these 8 megapixel consumer cameras. The Canon D30 even at a 3.4 CMOS is better than many of those 8MP's. Yet they are 8MP vs 3MP. Again, res is only one factor.
Click to expand...

Not a good comparison. The sensor sizes are not the same. The optics are not the same. The DSP is not the same. The consumer camera sensor uses $20 of tiny glass, is probably 10% of the area, and has to resolve too much for that area.

If we are talking about 1/3 inch sensors with lenses designed for 1/3 inch and similar DSP, I would take the higher resolution output any time! If you want HDTV or film output then resolution has to be as high as possible.

Btw, consumer cameras are definetely not bad. I have a 5megapixel 2yo $250 HP945. It has an 8x zoom lens system and a single 2/3inch sensor, but it is still SUPERIOR to ANY affordable HD cameras I have seen from an optics point of view. CA is about 1/5 of what it is on my JVC or the Canon! And it resolves more than 900lines easily when cropped to 16:9 and scaled to 1920x1080 at ANY f stop and ANY zoom setting and in JPEG format. CA is about 1.5 pixel in worst case and that's only in the corner. Contrast is very good and uniform from center to edge at all settings, something I can't say about the JVC with its grey low contrast corners. The MTF must be night and day compared to the lens of these 1/3inch camera. Strange isn't it? :) 1/3inch lens must be very hard to engineer.
 
Barry_Green said:
Only in the CineGamma modes. This is a known issue from back to the original DVX100 -- CineGamma has no "knee" circuit, so you always want to protect against blowouts.
Click to expand...

Interesting, I didn't know that... Why is this, Barry?

Edit: I assume it still does have a knee circuit, but that it is preset for the CineGamma mode and is not alterable? This would make some sense, but still I'd like to have control over it. What about the blacks (stretch, press)?

Matt
 
It's the way they designed it. CineGamma gives you a good half-stop more exposure latitude; it's designed along the same kind of principles as FILM REC in the VariCam. But there's no knee at all, so when you hit overexposure things turn ugly quickly.
 
I thought tha cinelike D had a non-adjustable luma knee ?

Working with cinelike gamma is a technique by itself.

If you want it to bring the best, you have to shoot 0 master ped, and lower the black level level a little in post. You'll then "see" in shadows, even if you lower the black level in post.
It will look 200% better than shooting -5 master ped. (It crushes the lower black only).
 
BobDiaz said:
FIRST READ THIS LINK: http://www.tvtechnology.com/features/Masked-Engineer/f_mario_orazio-08.17.05.shtml
IF the Panasonic or any other camera was set to F5.6 or higher, the resolution is going to be less than if the camera was set to F2.8!!!!!
If anyone must shoot in bright light, use the 1/8 or 1/64 Neutral-density filters built into the camera.
Click to expand...
Getting the most resolution with F2.8 of course depends on pixel size and that is the biggest secret here again. Mario's example was based on a 1/3" chip that has 1080 vert pixels. And putting more glass between imager and the chart (like nd filters) of course blurs the picture a bit more.

Now, if anybody who has hvx, resolution chart and software to measure chart picture's MTF could measure the resolving power with constant MTF (I'd suggest that 20% MTF would be closest to real life) and with different apertures, that would tell us the real resolution of the chips.
This is possible of course only if lense's resolving power is good enough to show when airy disk phenomena starts reducing the resolution.
It would be most accurate if the light would be very monocromatic eg. with only one wavelength. This could be done by using heavy color gels in front of the lights.

Let's say that we end up using 510nm green light.
If the resolution starts to decrease with F4, then the airy disc size is 0.51x4x1.22= 2.49µm which leads to vert rez of 1080.
If decrease starts at F5.6 then chips vert rez should be about 840.
If decrease starts at F8 then chips vert rez should be about 590.

So far the estimations have been that hvx would have a vert rez of 640. If you take out pixel shift effect and add kell factor, you would end up with about 600 vert pixels in the chip. With this pixel size airy disk effect wouldn't have any effect in F5.6.
Maybe hvx really has pal16:9 chips (1024x576)...
 
JanBee said:
With the new Sony XDCAM-HD PDW-F330 to arrive in March, we're going to have a cam surpassing all prosumer cams here by far - at a reasonable price tag of $13K (1/2" lens included!).
Click to expand...
And PDW-F350 might became a quite of varicam killer. 1(or 4)-60fps but in 1080p!
And PQ should be better with 35Mbps interframe than 40Mbps intraframe.
 
John_Hudson said:
After 17 pages of replies it's time I threw in the my wrench

The mojo. The magic vibe. Surely one camera has it over the others ?

The DVX100 has the mysterious mojo that everyone see's; the 24p, the latitude, the contrasts, something.............................................

Who cares what camera has the sharpest rez; isn't that negilble at best anyway's ?

What camera has the mojo ?

No, the XL2 cannot be tweaked in post to look like the DVX100 and forget about the FH and FX's and the XKLL100's.....

So tell us; wat camera has the mojo magic ?
Click to expand...

AMEN!!!!!
 
Barry, I've always wondered exactly how you get more latitude with cinegamma. Wouldn't a knee give you more exposure latitude due to compressed highlights?
 
Yes, but cinegamma is even more compressed. You can show more highlight detail in cine-D than you can in HD Norm with the knee set on "low". What you can't do is overexpose cine-D.
 
I just returned from a shoot in Kobe, Japan. Since I've had an HVX on order for quite some time, I took the opportunity of two free days to catch up with a Pana engineer, now based in Osaka, who I befriended while he was assigned to Pana's U.S. Pro business a few years ago. Though specifics were a bit thin and I'm not an engineer, it was possible to piece together information that seemingly brings insight to the way the HVX works.

From the insights, it seems the HVX engineers were quite clever in the way they derive HD from surprisingly modest components. It's understandable why Pana was reluctant to divulge certain specs. The insights would seem to explain several characteristics noted in the four-camera shoot:

The HVX's apparent superior sensitivity
The seemingly disappointing vertical resolution
The scaling artifacts in motion resolution test clips
The HVX's low basic cost

I'm in the middle of another U.S. shoot this week, so it will take a few days to assemble the info into a coherant form. Will post separately at that time. In the meantime, perhaps by now there are others who have pieced together insights into a complete picture.

By the way, I still have an HVX on order, though the HVX's reality seems a bit differeent now than when I ordered it.

More, soon.
 
You must log in or register to reply here.
Back
Top