View Full Version : Review on DVFilm's book, Shooting Digital


Heath McKnight
October 18th, 2004, 09:56 PM
Here's a review on the book by DVFilm to shoot movies digitally and possibly transfer to film (http://www.dvinfo.net/articles/filmlook/mcknight1.php).

heath

Simon Wyndham
October 19th, 2004, 09:50 AM
I quite like the book. Although I have found that it could do with more detail in places.

It is suggested that if you want your DV transferred to 35mm not to add any film look etc. Although it doesn't not explain quite how far one should go with picture grading to get a particular look. Obviously getting the transfer house to do this will cost a lot more.

Mind you, how many people actually have their DV movie transferred to film?

Heath McKnight
October 19th, 2004, 01:46 PM
No need to go for film look if you're going to 35mm, because you'll get it if you shot your film well, with a 35mm transfer in mind.

heath

Simon Wyndham
October 19th, 2004, 02:00 PM
True. But what if I want my film to have a particular look. For example bleach bypass, or a similar look to Three Kings.

Heath McKnight
October 19th, 2004, 02:08 PM
Hmmm, that's a good question, actually.

heath

Simon Wyndham
October 19th, 2004, 02:25 PM
Another problem that I have with the book is the constant recommendation of DVfilmmaker software.

Now, to give Marcus his due, he does outline all methods including 24p, frame mode on the XM (GL)1 and 2 etc. But to be honest I wish people would be honest about deinterlacing software.

DVfilmmaker for example, by his own admission, deinterlaces by interpolation on moving objects, thus halving the resolution at those points. Now imagine that you have moving camerwork like most feature films. That means that you are only getting half resolution picture and not the deinterlacing without loss of resolution as is often claimed.

The only way to obtain progressive scan at full resolution is to shoot that way to begin with.

Further, considering his experience I am amazed that he points out the 3:1 contrast ration limitation of video. I can feign ignorance as I'm not a technical guru (but I'm learning ;-) ), but I am surprised he hasn't got knowledge to the contrary as per this BBC white paper's conclusions;
http://www.bbc.co.uk/rd/pubs/whp/whp-pdf-files/WHP053.pdf

This points out via a full technical explanation that video is nowhere near as limited in lattitude as is often said. It even dissects the reasons why this myth seems to have come into being. It makes for interesting reading.

Marcus van Bavel
October 19th, 2004, 11:38 PM
<<<-- Originally posted by Simon Wyndham : DVfilmmaker for example, by his own admission, deinterlaces by interpolation on moving objects, thus halving the resolution at those points. Now imagine that you have moving camerwork like most feature films. That means that you are only getting half resolution picture and not the deinterlacing without loss of resolution as is often claimed.
-->>>

Sharpness or resolution is not perceived well for moving objects, only for relatively static objects, where the interpolation is suppressed.

Also the book recommends that you not constantly move the camera if you're shooting interlaced video for transfer to film, but use tripods, car mounts, etc.

<<<--
Further, considering his experience I am amazed that he points out the 3:1 contrast ration limitation of video.
->>>

No it never says video is limited to 3:1. It does recommend a lighting ratio of 3:1 as a goal, with the understanding that the actual ratio will probably be higher. Chapter 1 states the limit as 4 stops (16:1) which is valid for most low-cost cameras, and the accepted value is 30:1 (about 5 stops) for pro cameras as it mentions in the BBC article.

The BBC article mentions but largely ignores color resolution issues when you get down to RGB levels of 16 and less. For example, if you shoot 4 stops underexposed on DV you will only have 4 bits of color, which results in banding, etc. not at all present in filmed material that is underexposed by the same amount. The color performance limits the dynamic range as well as noise and sensitivity. The article also slights the problem of noise, which is much higher in prosumer cameras than the HD cameras tested there. To get the claimed dynamic range of 1000:1 for example would require not only 10+ bits per channel but a noise level of -60db and that is just impossible for a camera that costs less than ten grand.

Here is the link to the book: http://dvfilm.com/book.htm

Simon Wyndham
October 21st, 2004, 05:16 AM
>>Sharpness or resolution is not perceived well for moving >>objects, only for relatively static objects, where the >>interpolation is suppressed.

True. So given that this is the case surely one way of covering up for videos lack of resolution would be to keep the camera moving?

>>Also the book recommends that you not constantly move the >>camera if you're shooting interlaced video for transfer to film, >>but use tripods, car mounts, etc.

Hmm. I haven't read the absolute latest edition of the book, but are you saying for interlace transfer that the camera should be more static, but for 24/25p that camera movement is fine (accounting for general practises regarding film camera movement)?

When only a moving object is deinterlaced and the static background is left alone I always feel there is a strange look as if the moving object has been pasted onto the background. It's hard to describe, but I saw a BBC programme once (Little Britain in ine of the Prime Minister sketches for anyone interested) where it looked as if they had done something like this.

>>the accepted value is 30:1 (about 5 stops) for pro cameras >>as it mentions in the BBC article.

Okay.

>>example, if you shoot 4 stops underexposed on DV you will >>only have 4 bits of color, which results in banding, etc. not at

Yes the BBC article does have limits, though it does point out that generally it was intended for high def cameras.

>>channel but a noise level of -60db and that is just impossible >>for a camera that costs less than ten grand.

Okay.