Anyone done a short with the HD100U yet (or long form tests), and if so, is it posted or viewing? If so could you provide the link?

Here's an extended low-light test. There is another thread somewhere all about it.

http://homepage.mac.com/timdashwood/.Movies/Ex-lover_test_shoot_assembly-sorenson.mov

Here's an extended low-light test. There is another thread somewhere all about it.

http://homepage.mac.com/timdashwood/.Movies/Ex-lover_test_shoot_assembly-sorenson.mov

What did you shoot with? The Fuji lens? Any insigts as to your settings. I assume you had the gain all the way up. Did you use any adapters, such as the Mini35 or Micro 35?

And how did you get the slowmo effect?

Thanks

Duke

There really is another thread that answers all of those questions somewhere - I just can't find it right now.

Anyway, here are a few of the specs:


-shot between 9PM and Midnight in the rain. (no dropouts by the way.)

-Ambient light level was around 2 footcandles - 5 in the hotspots of the street lights. I used a 60W china ball for the CU handheld running shot.

-Stock lens with a Promist1 - ƒ1.4.

-0dB gain, 1/48th shutter, preset 3200K.

-MAX Standard gamma, black level NORMAL, black stretch 3, Knee at 85%, cine matrix (but the edited sequence has been colour corrected.)

-All shots were 720P24 except for the slo-mo which was 576P50, conformed in cinema tools to 23.98fps, then uprezzed to 720P (and then the edited sequence was compressed into a small file size for the internet.)


The point of this test was to see the absolute limits of the camera in the worst possible available light situation on a rainy night. I have so far learned a great deal from this particular test about how sensitive the HD100 actually can be (contrary to popular belief.) The progressive CCDs themselves may be less sensitive than their interlaced counterparts, but with proper control over camera processing (cranking the gamma and stretching the blacks) I found that it is actually VERY sensitive - even at 0db gain. I was able to show in another test that it was 1.5 stops more sensitive than the DVX100 using my "low-light" scene file, which would make it at least 3 stops more sensitive than the Z1 (based on Barry Green's shootout test http://dvxuser.com/articles/shoot3/ )
Based on what my lightmeter was reading (4-5 footcandle key) the night of the shoot, I would rate this gamma curve between 650-800ASA.

There was some intermittent SSE on three of the shots. It can be easily fixed in post by either crushing the blacks or making a slight black adjustment to one side of the other. I have also since found another way to eliminate it from the camera while shooting with this setting and a master black level of -2 or -3, but there would be a slight loss of detail in the blacks.

This footage is being made into a teaser trailer for a short film I will be shooting in a few months, so I'll post that when it is done.

Tim

Thanks Tim. I am looking for as much info as I can find about this camera going through its paces. Folks if there are additional links please post it. There should be a central locations where links for downloadable footage testing this (and other) cams can be found on this board.

-All shots were 720P24 except for the slo-mo which was 576P50, conformed in cinema tools to 23.98fps, then uprezzed to 720P (and then the edited sequence was compressed into a small file size for the internet.)
[/b]


So, in order to do overcranked shots, you need to downconvert basically? I am thinking about using the HD-100 for an indie feature, but I would hate to lose resolution in order to get slo-mo shots.

So, in order to do overcranked shots, you need to downconvert basically? I am thinking about using the HD-100 for an indie feature, but I would hate to lose resolution in order to get slo-mo shots.

If your project is 720p HDV, and you want slo-mo, you'll have to do with SD res cut in with your 720p.

If your project is SD to begin with, then you lose no res and it's win-win.

If your project is 720p HDV, and you want slo-mo, you'll have to do with SD res cut in with your 720p.

If your project is SD to begin with, then you lose no res and it's win-win.

Yeah, but who wants SD footage in the middle of an HD feature? Any chance this problem will be solved in the coming months?

It's not really a resolvable "problem." You can't crank video cameras. They either do a framerate or they don't. My suggestion is to shoot the spacial resolution you want, and then use your NLE to stretch it. Some software is *very* good at interpolating motion and creating the tweens.

Human visual perception is much more sensitive to spacial rez than it is to temporal.

It's not really a resolvable "problem." You can't crank video cameras. They either do a framerate or they don't. My suggestion is to shoot the spacial resolution you want, and then use your NLE to stretch it. Some software is *very* good at interpolating motion and creating the tweens.

Human visual perception is much more sensitive to spacial rez than it is to temporal.

The camera can shoot HD at 60 fps, right? Why can't there be a software solution that changes it to 60p for slo-mo, at full res?

Yeah, but who wants SD footage in the middle of an HD feature? Any chance this problem will be solved in the coming months?

Yeah it's called the GY-HD7000. It will be out in 2006. Shoots and records 720P60 at a much higher bit rate with a 2/3" CCD. $27000 without a lens though. Soon the HVX200 will be available and it will be capable of overcranking to 60P in HD. It will also do some other frame rates in between as well. The other option right now is the Panasonic Varicam 960x720P.

I don't consider this to be a "problem to be solved" in the HD100. JVC is the first to even offer 576P50 or 480P60. You should be glad they threw it in in the first place. I would love a 720P60 capture/record capability to be "turned on" with a firmware update, but if we are still limited to 19.2Mb/sec, then I don't know if the image quality would be worth it. Remember that JVC has optimized the MPEG2 encoder to give data priority to 24 frames within the 19.2Mb/s, so the other 36 pulldown frames are not as high quality. If we were required to "see" every one of the 60 frames, the 19.2Mb bandwidth would need to be evenly distributed to all frames.

The camera can shoot HD at 60 fps, right?
No it can't. It records 720P24, P25, and P30 in a 720P60 "wrapper."

If you've ever seen what Twixtor can do, you'd have little concern about overcranking in camera.

If you've ever seen what Twixtor can do, you'd have little concern about overcranking in camera.

Twixtor is amazing, but there is rendering involved, and it still doesn't capture the nuances of motion captured by overcranking.

No doubt, but I'd take it over upsampled SD for most normal "dramatic" uses. If you're trying to capture a bullet splitting an apple, that's a different story. ;-)

Twixtor is amazing, but there is rendering involved, and it still doesn't capture the nuances of motion captured by overcranking.

What is Twixtor?

What is Twixtor?

http://www.revisionfx.com/rstwixtor.htm

cant you capture 720 60p from the component out? I thought that was possible maybe I'm wrong.

The stream is indeed 60p, but it is made up of at most 30 actual frames (in that case, each one recorded twice.)

The stream is indeed 60p, but it is made up of at most 30 actual frames (in that case, each one recorded twice.)

No. When shooting in 30/60 mode, the component outs are 60fps. Only 30fps gets recorded to tape, however

I don't think people should dismiss the 60p SD mode. It is true widescreen SD not DV which has a lower effective pixel count. It up rez's very well to 720p and gives you some options in post. Including further slo-mo with software. While software can do a nice job for 720p, it does take quite a bit of rendering and isn't perfect.
I use this 480p60 mode often on the HD10, and after we output to DVD no one can tell that some shots were done at a lower rez. I have been asked how I got such a sweet slo-mo though ;>)

The "SilverOak" device (described in another thread) is gonna be awesome for this very purpose if it can capture that 60p without a huge video village or laptop or whatever. It will also be great for fx because of the better colorspace.

Thanks for the correction ... I didn't read carefully enough to catch the "component out" part of the question.

And sure, if the end-product is a DVD then you should have no concerns about shooting SD for slow-mo. My mind is "film, film, film" so the loss of spacial rez would be unacceptable.

There are certain techniques, such as up-rezing and sharpening, that will help to bring the footage up to very close to 720p image wise.
Even if your mind is "film, film, film" you must comprehend that many features have made it to the big screen that used DV (340,000 effective pixels) while the HD100 offers true wide screen SD (460,000 effective pixels) @60p. Of course the HD100 does have DV mode as well but you wouldn't use it unless a project had to use that codec.

Well, again, we are much more attuned to spacial rez than temporal rez ... that's why are brains are willing to be tricked into processing 24P as a moving image.

I know DP's like to do everything possible in camera, but given the choice between more pixels and less frames, I'll pick more pixels almost every time. It sucks to have to pick, but since we do, it's easier to fool the eye with interpolated slo-mo than it is to manufacturer pixels.

P.S. The claim that "many" films have been shot at SD is a bit disingenuous since any notable example does not even come close to looking like it is film. (Blair Witch, 28 Days, whatever.) They look like DV, and were shot with that in mind. Nobody tried to pass them off as anything other than what they are.

About the slomo shots in SD... some of The Matrix bullet-time sequences were shot using dozens of PD150s, so I don't see any problem there.

I'm sure there will be plenty of other things you'll have to worry more about in your prodution then the fact your 5 seconds of slomo are 480p60.

About the slomo shots in SD... some of The Matrix bullet-time sequences were shot using dozens of PD150s, so I don't see any problem there.

I'm sure there will be plenty of other things you'll have to worry more about in your prodution then the fact your 5 seconds of slomo are 480p60.

I thought these were filmed with still cameras?

http://whatisthematrix.warnerbros.com/img/stills_popup4.jpg

Yeah, I am concerned about downconverting the slo-mo, because I want a picture that matches, all in HD, in case the movie gets picked up for distribution (35mm blow up) and for HD-DVDs.

I might not remember exactly how the "bullet-time" effect was created for the first installment, but I know for a fact that the later installments used hi-res digital cameras (still) for those shots.

I highly doubt that an SD camera was used for anything other than pre-viz, or maybe as highly manipulated raw material to fill in tiny gaps ... but it's been a long time since I read up on it, so I can't speak definitively.

All I'm suggesting is use your common sense. If the human brain will accept a crappy frame rate of 24 fps as "motion" but we need 35mm resolution to project height and width, then it is obvious that the spacial rez is more important than the temporal. This is just a fact.

So if you have to fudge something, more often than not it would be the slo-mo, and like I've said, there's some pretty incredible post solutions for doing just that.

And there is some pretty incredible post solutions for up-rezing.
Few if any casual movie go'ers have a clue that Open Water, 28 days or Jackass the movie were DV/SD based. The studio's have some amazing kit when it comes to processing low rez footage. All did very well box-office wise. You wil notice that most slo-mo is usually very soft, but recently the Matix films and others such as Swordfish have given us the ultra sharp SFX slo-mo. Do not believe you will achieve such effect from 720p24 with software slo-mo. Not going to happen!

Higher rez and higher framerate. Both have their advantages. Here is the conundrum. You shoot at 720p24. You double that in post to 48fps. That is only a 50% slo-mo. The software should be able to handle it reasonably well. But, it's not that slow, and it will be softer. Now if you want to go slower, lets say 5X to 120fps, you will have a very soft image that will be prone to artifacts. At this point I would claim that SD60p up-rezed/sharpened then doubled only 2X to 120fps would give you a much cleaner image.
There is no "easy road" as both solutions require post software knowledge. Although SDp60 does 50% slo-mo right out of the box and it looks pretty sweet.

The 480p60 looks really good and is the same as a 50% time warp on the footage. The difference is that it is ultra smooth compaired to an NLE timewarp. With the footage already slowed down to half time in the camera you can get away with murder in your NLE because a 50% timewarp on already half speed video net's you 25% of actual real life shot speed. That is painfully slow.