skibum5
Offline
• • • • • •
Upload & Sell: Off
|
 p.2 #20 · Don't you dare modifying the 1DX, or else! | |
dswiger wrote:
This kinds of threads make me laugh & grimace at the same time.
I can assure you that the 1DC is not identical to the 1DX hardware wise.
The sensors may be sorted for thermal properties, maybe the process tweaked.
Maybe the ASIC has been rev'd to manage the DMA of sensor data at higher rates.
This level of design & test is not free.
I can also assure you that software is NOT free. The point being that every line of code requires extensive functional & regression testing.
While you "might" be able to juice up a 1DX to do 4k video, there very well may be buffer, cpu & heating issues. The job of Canon, if they do it right, is to guarantee reliability & performance.
The same is true in the auto industry. Yea, you can mess with advance curves, fuel delivery, even boost limits with aftermarket chips & tools. But if you hole a piston or some other catastrophic result, your warranty is void, and it should be.
If your camera locks ups or worse, you toast the sensor (it's not just heat sinks), then sending it to Canon is not kosher.
I have about 30 years in the firmware biz as well as testing methodologies so I know of what I speak.
I am also a software "tinkerer" so I appreciate what you are after. Just don't simplify it so much,, it makes me laugh & grimace at the same time & it ain't pretty 
Dan...Show more →
Or maybe not. The fact that they may so fear it implies that hardware and everything is pretty much all there, other than the heat sink. If the ASICs in it are not fast enough, then what on Earth would they be afraid of? (just for starters)
As for the 4k stuff, depending upon how they are getting 2k out of it there is no saying it was that much work necessarily. It's hard to know for sure. Since the 1DX doesn't have the a magic sensor size and yet since people seem to claim it has no moire/aliasing and is crisper than the 5D3 it almost sounds like they are maybe doing full reads on it and then all it is down to is what scaling factor you apply to 2k or to 4k and how many bits you have to push through certain stages per second. It might be a trivial change. Although perhaps not, granted.
The early talk from them was that it came at a high price and had required extensive re-design of the circuit board layouts and body housing and such to handle 4k and then all of sudden someone cracks an early one open and sees zero difference and now someone cracks one open and see only a single heat sink difference. Assuming rumor reports are all true. You do wonder a bit why they, assuming again reports are true, they had to make up all that hooey....
Whatever the case, it is doubtful it would have raised 1DX price by 1000-2000, if so, then how did the 5D2 come in under the 5D price despite it needing all the video work done, ALL, from the ground up, etc.? I always see people claim that adding any feature, even a working AutoISO, etc. to firmware would take years and costs hundreds more per body, which, if true, would mean a Rebel would need to cost $20,000. Not that things might not be trickier than some think or more costly, but....
A large part of the cost might have been in making the whole C100/C300 system and setting up all that stuff and the new cinema team and they can't have a 1DC go putting that all to waste, it has to be priced very high and the cost is not so much to cover DSLR stuff but to proctect money put into the other stuff. Maybe pricing it lower makes them take a bath on that. I was sort of figuring the whole C100 thing kinda meant we wouldn't get non-stop DSLR video revolutions any more, not sure you can really feasible have it both ways. Part of me thinks it would've been far better to forget doing all that (and I have seen similar comments from some big film guys) and just pop out the best video you can on your DSLRs, that was a truly new segment, one to create dominate for a while, but who knows, maybe they know what they are doing even if it is not so good for us types.
(I'm not an embedded firmware engineer, but I have coded to custom chipsets at the hardware register level before plenty of times. I have coded in everything from BASIC to 680x0 assembler to Action to Perl to PASCAL to Fortran to C to Object-C to C++ to etc. and at the hardware register level or using upper level APIs on a number of different platforms. And not that I want to trivialize programming, it can take a lot of time and effort to do many things, no doubt.)
Anyway, it is there right, of course, to do what they want, and tier and segment as they wish. They prefer a safe, conservative, just one of the players approach and than a segment revolution. Maybe that was very wise or maybe foolish.
Edited on Jan 09, 2013 at 06:27 PM · View previous versions
|
