Sun Jun 16, 2013 1:05 am
I suppose it is possible future generations could come in a HS model-and it would be nice if the spectral response curve was identical to the sync speed camera (it would fit traditional model of cameras- the 435 and 535 both came in standard and HS models. Even an Extreme model). That would be nice-you don't always need HS footage, and switching cameras on the few days you need it, it would be nice to have a camera you know will cut well-rather than having to test FS700 vs Phantom vs whatever else you want to try.
What I think is very promising with this chip is its adjustable Dynamic Range mode. If you go into the product flyer, it is able to adjust its dynamic response (sort of similar to a knee adjustment) on the fly-on the chip. Knee adjustments traditionally is for compressing a 10-stop range say into an 8 bit recording medium. It did nothing to increase the size of the "well" of the actual chip.
This sensor does this on the fly-on the chip. Possibly with the global shutter circuit. If you look at the specs, the actual native dynamic range is 10-stops (13e- to 13500e- = 1048, about 10 stops), so BMD must already be doing some of this to get to their 12stop spec. The chip maxes out at 15 stops (however will only quantize 12 bits, so to get to 15 stops you necessarily get less granularity between light values.)
It seems the 12 stop spec is to get a good balance between DR and no banding (mathematically, 12 stops requires 12 bits-due to the similar exponential nature of light and digital quantization) What would be interesting is if BMD can develop a couple extra curves, similar to Sony with their S-log, and offer a wider range of DR to customers in the existing camera. The trade off being higher DR/less latitude to grade in post, or full grad-ability they have now/slightly less DR. The artistic decision would likely be based on how contrasty of an image you want to have (more contrasty, go for more grade-ability. If you want a low-con look, start with less contrast and less ability to dial the contrast up.)
Looking at the quick specs it is entirely doable in the 4K camera they are coming out with now, with no hit to on-board CPU resources, data throughput at any point in the pipeline. It would mean extra metadata, and extra color science and possibly an extra mode in resolve to handle the new log curves. but totally doable with nothing more than a software change.
One other standout from the product flyer-it has a 2 sided anti-reflective coated glass lid...which means there should be no shenanigans with one supplying the chip who then hands it off to a glass bonding who then messes up the quality. They have been attaching this glass successfully for the last year, and so hopefully they will not ship any sensors that don't pass BMDs QC. It should be a simple thing of the assembling robot taking the sensor out of the tray, placing it on the board, send it through the soldering oven and presto. Should lead to a high yield production line (perhaps all of these weighed into BMD's thought process in selecting a new chip for their 4K camera?)
Michael Collier, II
Cinematographer - Gaffer - Camera Operator
http://www.RandomAcronym.com
http://www.imdb.com/name/nm3075740/?ref_=fn_al_nm_1
