Our Hands-On HD100 / Mini35 Test: Articles, Photos and HD Video

[Chris Hurd]

The results are in, see http://www.hdvinfo.net/articles/jvcp...ini35test1.php.

[John Jay]

'twas an interesting read,

However I'm not surprised with the diffused soft image look produced by this setup considering the high quality components in the optical path.

The effect of an oscillating ground glass is to soften and diffuse the image considerably more than a fixed static ground glass and I suspect there is a point at which you cannot get any more juice out of the melon.

I would doubt the wisdom of putting a 'HD whatever cam' behind a moving ground glass unless you just happen to want a big image and not more detail.

[Nate Weaver]

Sometimes the image coming from the Mini35 seemed soft, but I personally think it's a little decieving.

Comparing the 16x and Mini rig both on set and by looking at the clips later, I'd say they're very close.

[Charles Papert]

Originally Posted by John Jay

I would doubt the wisdom of putting a 'HD whatever cam' behind a moving ground glass unless you just happen to want a big image and not more detail.

John, can you elaborate on this? I would think it means the same thing--if you can project a bigger image, doesn't that mean that you have more information, which naturally means that more detail is present?

To me, digitally projected HD features can sometimes appear too sharp and detailed. My feeling is that this should be an option, but perhaps sometimes a more muted look is desirable for given material, much in the same way that different film stocks or older series lenses are selected for film-originated material.

[John Jay]

Originally Posted by Charles Papert

John, can you elaborate on this? I would think it means the same thing--if you can project a bigger image, doesn't that mean that you have more information, which naturally means that more detail is present?

Charles, no this is not what I'm saying. --- It would appear that the diffusive and bluring effect of the moving ground glass is capping the resolution in the image so although you are shooting 720p you are effectively recording an SD image into 720p space.

For example in the CU shot of your GF, the Bokeh of the background vegetation appears sluggish or perhaps posterised. When I first played this clip I was about to think it was shot with something like a 300mm, and was surprised to learn it was only 100mm. The reason I thought this is because it looks flat without the 'figure and ground' standout nature such a shot would create - in short to be HD, her outline should be crisp and pop out against the background but alas it just looks to me like a texture difference and as mentioned earlier it is the diffusion which is most likely the cause.

[Stephen Finton]

Originally Posted by Charles Papert

John, can you elaborate on this? I would think it means the same thing--if you can project a bigger image, doesn't that mean that you have more information, which naturally means that more detail is present?

To me, digitally projected HD features can sometimes appear too sharp and detailed. My feeling is that this should be an option, but perhaps sometimes a more muted look is desirable for given material, much in the same way that different film stocks or older series lenses are selected for film-originated material.

Wouldn't it be the square pixels that make HD look too sharp, compared to film?

[Barry Green]

SD on a 720p frame? I think that's a mischaracterization. You're saying that the image has no more resolution than an SD frame, just spread over the size of a 720p frame? That would require the HD100's 720p frame to be blurred by a factor of 3:1 in order for the results to end up at standard-definition quality.

I think there may indeed be a little bit of softening of the image. It only stands to reason -- we're shooting through a piece of ground glass, after all. But comparing the subject and the detail between the mini35 version and the Fujinon version, it's nowhere *near* the level I'd expect from standard definition.

in short to be HD, her outline should be crisp and pop out against the background but alas it just looks to me like a texture difference and as mentioned earlier it is the diffusion which is most likely the cause.

I'm not so sure about that. I think the reason is an overall even level of tones (and a corresponding lack of contrast) in the shot. We discussed putting a backlight edge on her, but we liked the shot and the composition the way it was. Had we put that edge light in there, it would have provided a 100+ IRE highlight in the frame tracing her shoulder, which would give the scene a lot more contrast, a lot more "snap". That ultimately was not the look we decided to go for, although we did do that for the night shots.

I refer you back to the A-closeup.jpg shot, where we did have a small side edge light -- she just leaps off the screen. Sure the background is blurrier than in the flowerbed, as to be expected from a more close-up shot -- but the main difference is that we perceive the bright sidelight edge, and its attendant contrast, and it gives great separation from the background. And that shot looks plenty sharp, even having been through the diffusion of the ground glass.

[John Jay]

SD on a 720p frame? I think that's a mischaracterization. You're saying that the image has no more resolution than an SD frame, just spread over the size of a 720p frame? That would require the HD100's 720p frame to be blurred by a factor of 3:1 in order for the results to end up at standard-definition quality.

I think there may indeed be a little bit of softening of the image. It only stands to reason -- we're shooting through a piece of ground glass, after all. But comparing the subject and the detail between the mini35 version and the Fujinon version, it's nowhere *near* the level I'd expect from standard definition.

Barry forgive me for saying this but it seems like you are always prepared for these tests and it surprises me that you did not have a resolution chart somewhere in your gunny sack with which you could have performed a resolution test and removed all doubt ( maybe you did perform such test but are keeping a lid on it?)

In any case heres some math to consider;

The stationary Arri Ground Glass (GG) is approximately 20 micron guage,which, on an academy gate width of 22mm gives a theoretical maximum resolution of 1100 lines or 550 line pairs. Factor in the diffusion angle of around 10 degrees and what is left is say 1000 lines. Now lets say the oscillating GG has an eccentricity of 1mm ---- thats +/- 0.5mm from a standing position. Now 1mm is 50 lines and so the diffusion/blurring characteristic is operating over 50 lines ( thats why the grain disappears). So, now its a case of 1000 line capacity with a 50 line diffusion. Now diffusion is not as severe as good old gaussian blur, but nevertheless does have a significant blurring component, which I would reasonably place at 500 lines, which is about what I am seeing in the clips. The optimal solution would be a PAL XL2 with Mini35 for this gate size.

Now if P+S should want to give us HD quality imagery, the solution would be to up the gate size to the 36mm width of Vista-V and we would end up with a respectable 800 lines. It is not unsurmountable - I guess they are too wicked and lazy :)

I refer you back to the A-closeup.jpg shot, where we did have a small side edge light -- she just leaps off the screen. Sure the background is blurrier than in the flowerbed, as to be expected from a more close-up shot -- but the main difference is that we perceive the bright sidelight edge, and its attendant contrast, and it gives great separation from the background. And that shot looks plenty sharp, even having been through the diffusion of the ground glass.

Whilst I would agree that some form of rim lighting would provide a visual clue to figure and ground seperation, it is not what I am referring to. Whether lit or otherwise, it is the edge definition which provides the perspective clues over image depth. Perceived definition falls off with increased viewing distance and conversely images with low edge definition are perceived to be closer to the background no matter what strength of Bokeh, thats why I say the movie clip containing the jpg you refer to looks flat or even woolly - the edge definition is not there and you need to be in the 800 line ball park to make the foregound pop or stand out from the backgound.



|E|DIT>>

check out Matteo Ricchetti's work in the Alternate Image forum for evidence of a 'real high definition' adapter , which uses a bigger GG. The clip of the dark haired girl close-up really is crisp and stands out a mile. His work is promising and is definitely in the right direction.

[Mathieu Ghekiere]

Very funny beginning of the article, like it is a fairytale :-D
btw: congratulations Charles, nice article!

[Chris Hurd]

Thanks Mathieu, the history of this business seems like a fairy tale at times.

[Will Griffith]

Any future plans to test the component output from this kind of setup?

If HDV looks this good I could only think the pre-HDV compressed output
would look even better...and actually be good for effects shots.

[Nate Weaver]

Originally Posted by John Jay

Barry forgive me for saying this but it seems like you are always prepared for these tests and it surprises me that you did not have a resolution chart somewhere in your gunny sack with which you could have performed a resolution test and removed all doubt ( maybe you did perform such test but are keeping a lid on it?)

John, I know you don't know me from Adam, or what equipment I've worked with over my career, or etc etc etc...

But trust me on this one: there's a lot more than SD resolution captured by the Mini/HD100 rig. :-)

I've been looking at HD images on a pixel-for-pixel display for over a month now every day, and while HDV images can definitely be underwhelming at times, I've learned that the fastest way to get your perspective back is to bring up some SD images on the same display.

In other words, I'm a little sorry about how fast my brain has gotten used to HD images...HD now looks like SD used to, and SD just looks deficient!

[John Jay]

I've been looking at HD images on a pixel-for-pixel display for over a month now every day, and while HDV images can definitely be underwhelming at times, I've learned that the fastest way to get your perspective back is to bring up some SD images on the same display.

In other words, I'm a little sorry about how fast my brain has gotten used to HD images...HD now looks like SD used to, and SD just looks deficient!

I suspect the pivotal objection here is to the term SD. In PAL land (homeground for me) 720 vrez is only 25% more whereas in NTSC land 720 is 50% more vrez.

Now you cannot dispute that firing photons through a multi element lens followed by a moving ground glass followed by a condenser field lens and then bounced through a prism and then through another lens and then split three ways through another prism is gonna knock that 720 down to something approching PAL SD? Please, trust your feelings?

Maybe its something more than NTSC SD, in which case I would totally agree with you.

[Charles Papert]

I have just spoken to my contact at JVC, who tells me that they have screened this footage on a number of displays, from LCD to HD CRT to front projection D-ILA...he says that the front projector, which is working at 720p native resolution, looks the best since it is not scaling; and the funky posterized look that we have noticed in the out-of-focus backgrounds is simply not present in this configuration. He says that projected on a 10 foot screen, the image is simply stunning.

[John Jay]

Charles,

That's good to know and thank you for the information :)