Spent some time sharing and exchanging ideas with fellow filmmakers, photographers and image makers - Ryan Hughes, Kar Wai Ng and Arash Moallemi. We did a little informal testing but mainly discussed workflow, shortcomings of DSLR cameras, as well as imaging technique and application in general. Hopefully, the first of more meetings to come.
Some pics from the day - all photos by Arash - thanks!
barrycheong.com 7D package next to a standard 5D.
Ryan testing out handheld with my 7D. He was very interested in exploring handheld shooting, as his last experience hadn’t been too favorable. The added hardware of my package definitely makes handheld easier to accomplish.
Kar Wai and I talking shop…
Click Above to View
I saw a quick test my friend Ryan Glover did with his GH1 and a Panasonic anamorphic adapter attachment designed originally for the DVX100 and it got me interested in researching the idea further. Some other people have been experimenting with similar adapters and DSLR video. These anamorphic adapters, originally designed to convert 4:3 video to 16:9, have since fallen off the radar, but used with 16:9 native chips will give you roughly 2.35:1 aspect ratio plus all the characteristics of using anamorphic lenses, most famously the blue vertical streaking. The adapter essentially squeezes the image vertically by a factor of 1.33x resulting in everything looking skinny but in post you stretch everything back out by that same factor which results in your wider aspect ratio.
The adapter I’m using is a Century Optics Digital Series lens with 58mm threads that I believe was originally used with a Sony PD-150 back in the day. The 58mm thread works well with Canon EF primes like their 85mm and 50mm 1.4 and with a simple stepping ring, to their wider lenses which have the smaller 52mm thread. I’ve been testing mainly with their 50mm 1.8 which is the only lens I have access to at the moment. It will unfortunately not work with their L series lenses.
(My 7D w/ Canon 50mm 1.8 and Century anamorphic lens attachment.)
Click on the screen grab at the top of the post to see a quick test I did. Shot in available light between 640 and 1250 ISO, between T/2.8 - 5.6, and anywhere between 7-15 feet roughly away from the subject. I flared the lens manually with an LED maglite for fun as well. Please excuse my silly friends, who thought it would be silly to dress up and be silly in front of the camera. I think they’re pretty silly (and awesome).
There definitely are some limitations with focus and exposure with this adapter, but as such force you to frame, block and generally work slightly differently. For example the lens starts performing a lot better once you stop down to T4 - 5.6 (as do real anamorphic lenses used with 35mm cameras). Also, with the 50mm at least, you have to maintain at least 4 or more feet away from your subject to get focus. I briefly tested the adapter on an 85mm 1.8 and found focus difficult to achieve even at deeper stops. Putting more distance between lens and subject will probably help but at that point defeats the purpose of using that long of a lens unless you really need the spacial representation of an 85.
In general this adapter is not for every project but for a certain look, for a certain project it could be really great. More thorough testing to come.
I was curious to know whether the 7D performs any better, noise-wise, in tungsten or daylight. I ran a quick test with a Kino Flo bulbed with both types of tubes and shot all the combinations. Camera white balance was set to either 3200k or 5500k manually.
Shooting stop was T/4.0 @ 320 ISO. I used my 24-70mm lens. I didn’t take note of the focal length. 24P @ 1/50th of a second shutter. I also shot tungsten light in daylight balance and daylight in tungsten balance and corrected in Photoshop quickly with the Color Balance option, to see how well you could recover your image.
Results below:
No apparent difference between tungsten or daylight in terms of noise. The images shot in their opposing color balance are definitely noisier but not much more so. They’re also a lot more saturated. I didn’t bother to compensate that factor.
The last test during my initial testing phase was testing a general over/under exposure on the scene. The scene was lit to T8 @ 500 ISO base. I didn’t want to start shifting the lights to achieve my exposure so the scene was lit to middle of the road on the lens to allow me enough stops on either end of the base to open/close my iris. The lens was the same Canon EF 24-70 f/2.8L.
To give a sense of the overall contrast of the scene - the black BG read about 4-5 stops under, the banana’s were about 1 stop over and everything else fell between +/- 0.5-0.75 stops from key.
I’ve posted screen caps below. Same deal - Apple Prores transcode and then screen cap TIFFS made from that in Final Cut Pro. 100% size.
The very top image is the neutral image at proper exposure. The image on the left is the graded version. The image to the right is the exact same image but ungraded so you can see the image that’s coming straight off the camera. The grading was done in Final Cut with the basic 3 way color corrector. You might be able to achieve better results with Apple’s Color or something more sophisticated but this gives a general idea. All images were balanced to the grey card with a waveform and then tweaked to balance the rest of the items in the frame. Mainly midtones and highlights were affected with some compensation to chroma since that gets affected as you move your levels up and down.
Some initial thoughts and notes:
- The image falls completely apart at 2 & 3 stops over and under.
- Over/under 1 stop is definitely usable. In the case of +1 the image noise improves quite dramatically - you can see it in the grey card. The midtones and the highlights are starting to blend a little - you can see the orange and the face of the red box have less definition compared to the neutral image. The -1 image is definitely more contrasty and noise is more apparent. Knowing this you might be able to create a certain look depending how you expose.
- Generally speaking I’m thinking that rating the camera anywhere between 1/3 to 2/3 slower (like Neg Film) will probably improve noise performance while maintaining some definition between midtone and highlight.
I next tested Canon’s High ISO Noise Reduction option found in their Custom Function II menu option. The camera gives you 4 options - standard (STN), low (LOW), strong (STR) and disable (DIS). The results tested at 3200 ISO are posted below.
There seems to be no improvement or variation between all 4 modes. I also tested this at 800 and 1600 ISO with similar results. It looks like this function of the camera might be relegated to the stills mode. Bummer.
Spent the day doing some preliminary tests with my new 7D. I tested ISO performance, Canon’s ISO noise reduction modes and a general over/under test. I’ll release all my results over several posts just to break it up.
The first testing was for ISO and noise performance. For the test I arranged some objects on a small table. I wanted something with lots of color, something with texture and black. I also threw in the defacto grey card. I’ll probably do a similar test when I can get my hands on a colored chip chart. The setting was lit with a 650 Arri Fresnel with some 216 on the doors. Base was T2.8 1/2 @ 100 ISO since that’s kind of an ideal stop I like to shoot at. The lens is a Canon EF 24-70 f/2.8 at 35mm focal length. Shutter was 1/50 and white balance set to 3200K. Recorded in 24P. I started at 100 ISO and went up from there - 200, 400, 500, 800, 1600 and 3200 were all tested. I’ve posted some screen grabs below. The files were all transcoded to Apple’s ProRes 422 HQ and then screen grabs where exported from Final Cut Pro as TIFF’s. They are at 100%.
Some initial thoughts and notes:
- I found that the image is still very workable at 800 ISO. Once you start getting to 1600 ISO and higher the noise becomes quite apparent though depending on how much you might resize the image and where you might be displaying the content you might be able to get away with using these higher ISO’s.
- 3200 is quite a bit more noisier than 1600. I would consider using 1600 if I were in a pinch and had to shoot. Considering it’s 1600 ISO I’m still quite impressed.
- Surprisingly 800 ISO doesn’t seem much noisier than 500 ISO and gains you an extra 2/3rd stop.
- Sharpness steadily decreases as you increase your ISO.
- The grey card in the 3200 test I think shifted somehow as it’s brighter than the rest of the other examples. I plan on testing this again with a broader range of contrast and textures.
Testing recap specs:
Camera: Canon 7D
Lens: Canon EF 24-70 f/2.8L USM @ 35mm
Record Media: Lexar 16gb 300x UDMA Compact Flash
Format: 1080/24P
WB: 3200K
Light: Base T2.8 1/2 @ 100 ISO, 1/50th shutter
High ISO setting: Standard
