Why the downer on the 1D C?

[dustylense]

3 TB harddrives for $190, right there on the shelf. $399 HP laptops. Moore's law is anything but broken.

Canon's pricing department, that's another story.

Turn them over and see where they were manufactured?
Nikon F4 (film) was about $3800-4000 when launched in 1988. What does Nikon charge for the D4? $6000. Prices go down quality goes up? July 2002, Nikon d100 6mp camera, price $2000. How much for the Nikon d800? $2,995. But to quote you, "prices go down, quality goes up".
Competition however can drive a market. THose best buy laptops, how many were there? 15? What FF HDSLR on the market captures 4k to CF card at APS-H size? If there were 15 of them, it wouldn't cost MSRP $15k. Gotta start somewhere.

 

[Cosimo Bullo]

Fair enough... we'll leave this unsettled.

Besides, I too give my mom (or sister) my old laptops - so we've got that in common, and everyone's happy.

 

[dustylense]

Fair enough... we'll leave this unsettled.

Besides, I too give my mom (or sister) my old laptops - so we've got that in common, and everyone's happy.

Well, I try to sell them it first

 

[StMad]

Kool. I'll go out and get my new porsche for $20,999 before tax and licensing.

If a company could make a profit selling a Porsche for $21k, they would. That's what's happening in the (video) camera business. There are no doubt a few reasons why the big players price the way they do. Though if a $3k raw camera (and all that follows) hurts their business, then they'll have to change their approach and pricing.

To these eyes, it seems feasible that Blackmagic will produce an Alexa quality camera within the next couple of years, at a fraction of the price.

 

[dustylense]

If a company could make a profit selling a Porsche for $21k, they would. That's what's happening in the (video) camera business. There are no doubt a few reasons why the big players price the way they do. Though if a $3k raw camera (and all that follows) hurts their business, then they'll have to change their approach and pricing.

To these eyes, it seems feasible that Blackmagic will produce an Alexa quality camera within the next couple of years, at a fraction of the price.

But they don't sell Porsche 911's for 21k, do they. They sell Boxsters (or did) at the price, but they weren't 911's.
I like the Black magic announcement. But it's really too early to tell if this cam will dent anything. I hear reports of horrible rolling shutter. As in the BMDC thread, someone quoted ""I played with it and was super excited about the idea of a RAW camera from Black Magic. And then I panned it. Not quickly... just panned it. And the rolling shutter made a vertical line into almost a 45 degree angle across the screen. That was the end of my interest. I would subjectively describe the rolling shutter as worse than an iphone's." If thats the case....

 

[Kholi]

But they don't sell Porsche 911's for 21k, do they. They sell Boxsters (or did) at the price, but they weren't 911's.
I like the Black magic announcement. But it's really too early to tell if this cam will dent anything. I hear reports of horrible rolling shutter. As in the BMDC thread, someone quoted ""I played with it and was super excited about the idea of a RAW camera from Black Magic. And then I panned it. Not quickly... just panned it. And the rolling shutter made a vertical line into almost a 45 degree angle across the screen. That was the end of my interest. I would subjectively describe the rolling shutter as worse than an iphone's." If thats the case....

#1600

It depends on what lens was on it.

All cameras in this price point have a skew of some kind. The longer the lens, the worse it appears. I saw shots of the 70-200 Canon on the stand. That's gunna skew for sure on a pan.

Yes the skew is there when you look for it. But I've gotten used to dealing with it on 5D's, GH2's and even on RED's. Even the Alexa has skew in the right conditions.

Right now I'm shooting a TV series with extreamly long lenses on Alexa. I've had a 150-600 with 2 2x doublers...thats 4X......thats a 600-2400mm lens....it skews when you do a slow pan....

It's great to have an 'ahh HA" and point out a fault with a camera, but really, it's no worse to my eye without empirical testing, than other cameras in this price range. If you're expecting better, then it isn't.

jb

The case is just that the camera isn't finalized yet, out yet, and barely tested I guess. I'll take John's word for it, same reason why you take Shane's word for his I'd reckon.

 

[StMad]

Sure. I was weaned on Photoshop 2 (circa 1992)...

Thanks...I appreciate you taking the time.

Consistent, complaint and repeatable...makes sense. It's one thing to fiddle and get a result you like...but if you're doing this for professional results, that approach just won't cut it.

 

[StMad]

But they don't sell Porsche 911's for 21k, do they.

Is car production and camera production analogous? I admit I can't speak on the subject (but I will )...but surely parts, design and manufacture of a car are driving costs in ways which are different to a camera? And this looks like being a mass production item. If so, profit margins don't need to so great.

Rolling shutter is linked to sensor readout, no? If they managed to get raw, prores, dnxhd, 10bit, 13 stops, small form factor, 2.5k, sdi etc etc into a camera for $2800 (and throw in resolve and ultrascope), why should speeding up sensor readout be a price breaker? (assuming it's a problem).

 

[StMad]

It'd be interesting to know how much more Blackmagic would need to charge to get us in Alexa territory. Larger sensor (and all that involves), design, faster sensor readout, and whatever else goes in to making the Alexa what it is. You'd imagine they'd sell less with it being at a higher price point. Ok, I'm way off topic now...

 

[hendosan]

Blackmagic can't just make a larger sensor camera - they don't manufacture or do R&D on sensors so they have to buy a sensor from a 3rd party, and as far as I know the big boys don't put their chips up for sale to other manufacturers.

 

[John Brawley]

It'd be interesting to know how much more Blackmagic would need to charge to get us in Alexa territory. Larger sensor (and all that involves), design, faster sensor readout, and whatever else goes in to making the Alexa what it is. You'd imagine they'd sell less with it being at a higher price point. Ok, I'm way off topic now...

Wrong thread.

but they'd have to charge a lot.

That Alexa sensor is special. And expensive. Very very expensive. There's no magic bullet here, and this is also I suspect why the Canon C300 / C500 etc are also more expensive. Speaking to the Sensor guys at BMD, there's no free lunch on sensors. (when it somes to skew and low light)

I'm using a couple of C300's on a TV series at the moment as companions to Alexa. Shooting in LOG mode you can get them to grade if you expose them correctly, but can still pick them pretty easily. But they are great for certain jobs which is why we use them. Shooting cars and shooting in very very low light.

For this they produce magnificent pictures....

I'd hate to pay for one though. To me they do seem overpriced, but they do a great job for the times I need them. I'm lucky enough to be on a production that can pay for them.

jb

 

[squig]

Well, I try to sell them it first

I use my macbooks until they die....they generally don't live very long.

I'm interested in looking closely at the 1D C as a potential A-cam on my feature (particularly if Shane's comments are on the money) but for 15 grand I wouldn't just go out and buy one without funding for the film in place. The 5D MKIII is good enough for just about everything else.

The Blackmagic camera may be a better option though.

 

[Simon Shasha]

Blackmagic can't just make a larger sensor camera - they don't manufacture or do R&D on sensors so they have to buy a sensor from a 3rd party, and as far as I know the big boys don't put their chips up for sale to other manufacturers.

There's plenty of other companies that make sensors other than "the big boys" - i.e. Cmosis.

 

[vladszp]

most expensive part of sensor is R&D .... when u start production thay all cost same (depends of the size of sensor)...how much they going to charge for it depends of demand...if u spend millions for R&D and going to sell only few thousands than price must be high ...but if is dslr sensor(who going to sell in 100000 or more) than u can go cheaper...

 

[Jay Birch]

I'm still unsure why everyone is up in arms about the price.

As far as I understand, there are (or will be) six 4K cameras, this year, under $100k (for a basic shooting setup):

Sony F65 - $85+
RED Epic - $40k+
RED One - $30k+
Canon C500 - $30k+ (?)
RED Scarlet - $16K+
Canon 1D-C - $13k+ (?)

Each camera has it's own +/- points and a price that seems to fit this relatively niche, professional market.

 

[trez]

Interesting points.
No matter how sophisticated is your filtering,
0 0
0 1
will be equivalent to
1 0
0 0
and
0 1
0 0
and
0 0
1 0
when downscaled. Because the filter needs to be more or less symmetrical.
So no. Not really 10 bit equivalent, even with smart filters.

Good point, but not true, either (mine was an approximation, so it wasn't true either, as you pointed out). Here's why:
Your example highlights the fact that filter must 'commute' (the layout of four pixels being averaged doesn't matter as long as their values are the same - reordering them must result in the same output value). I fully agree with that. However, your example is based on 1-bit oversampled data (each pixel either on or off). In such case:
- the range of possible pixel values in original data is 0-1, which is 1-bit
- the range of possible values in filtered data is 0-4 (0-1*4), which is close to 2-bit (slightly more) - we gain ~ 1 bit here
Now take 8-bit source data into account:
- the range of possible values in original data is 0-255 (8-bit)
- the range of possible values in filtered data is 0-1020, which is virtually 10-bits (0-1024) - we gain ~2 bits here.

It depends on bit-depth of source footage. The higher it is, the closer we get to the limit of 2-bit gain.

This non-uniform distribution that caters to human eye exists on ALL consumer and pro cameras. It is called (non-log) gamma.

True. As far as I know those ordinary gamma curves don't deviate significantly from 2.2 gamma curve because they need to be viewable on standard displays (the 2.2 gamma was once imposed by CRT transfer curve). You're right the sampling is also non-linear here - so thanks for pointing that out!

Yes, they may tweak it all they want and try to balance it. And in all fairness, that's all they can do: try to find a decent compromise to stuff it into 8 bits. But it stops being true LOG then and lacks the true exposure latitude options of a true LOG.

Agree, but I think those more sophisticated gammas, while suffering from the obvious flaw of being 'unviewable' without the LUT, succeed in distributing the light levels into digital word sub-ranges in more 'linear' fashion. By saying 'linear' I mean linear in relation to our eyes (and brain) transfer curve - we are not as sensitive to absolute light level differences in highlights as we are in shadows and mids, and those gammas make use of this fact. Also, being more or less LOG-based they still allow for coding higher DR. IMO 8-bit LOG is much better than 8-bit encoded with standard gammas and that's all I meant here

 

[cpc]

Good point, but not true, either (mine was an approximation, so it wasn't true either, as you pointed out). Here's why:
Your example highlights the fact that filter must 'commute' (the layout of four pixels being averaged doesn't matter as long as their values are the same - reordering them must result in the same output value). I fully agree with that. However, your example is based on 1-bit oversampled data (each pixel either on or off). In such case:
- the range of possible pixel values in original data is 0-1, which is 1-bit
- the range of possible values in filtered data is 0-4 (0-1*4), which is close to 2-bit (slightly more) - we gain ~ 1 bit here
Now take 8-bit source data into account:
- the range of possible values in original data is 0-255 (8-bit)
- the range of possible values in filtered data is 0-1020, which is virtually 10-bits (0-1024) - we gain ~2 bits here.

It depends on bit-depth of source footage. The higher it is, the closer we get to the limit of 2-bit gain.

Yes, my example was not quite relevant because, really, permutation is irrelevant.
Yes, one would gain some resolution in tonality on downscale if working in a higher bit depth in post. Calculating the average error, compared to 10-bit initial qunatization, is a typical numerical methods problem. And there is certainly some deviation from an initial 10-bit quantization.
Quick example, to illustrate that: a flat wall might be quantized to 511 in 10-bit for all four pixels in a 2x2 block. In 8-bit the 4 pixels would all be rounded up to 128, and when integrated together would yield 4*128 = 512.

True. As far as I know those ordinary gamma curves don't deviate significantly from 2.2 gamma curve because they need to be viewable on standard displays (the 2.2 gamma was once imposed by CRT transfer curve).

It is a common misconception that gamma encoding is connected to CRT characterstics.
Typical 2.2 gamma is entirely based on human perception under specific viewing conditions and on our higher resolution in the darker tones, and provides for a more effective distribution of values over these tones. Straight linear encoding would just not honor resolution in the darks. The fact that a typical CRT has a gamma of around 2.5 was merely a (happy) coincidence. But it is not a coincidence that middle gray is 18% gray (and not 50% gray). Again, this has nothing to do with screens, but only with human vision and the need to effectively encode a tonal range for reproduction (usually on a limited contrast display). It is simply a matter of allocating more values resolution in the more sensitive (to the eye) areas - the darks, compared to what a linear encoding would allot. Or to rephrase, gamma encoding serves the purpose of preserving tonal precision when quantization is required, based on human perception.

LOG gamma in the digital realm only exists to ensure (approximately) equal tonal resolution over the entire captured dymanic range so that exposure tweaking in post is viable. This is not how we see things. We need any digital signal to be converted back to linear space for display (that's what a monitor does; it expands the gamma back to linear, harvesting the benefits of preserved tonal resolution in the darks) so that we can see correct luminance and colors.
Even if you project a LOG-gamma image (no gamma display involved) it would still appear washed out. The projection also has limited contrast and needs some contrast applied, even in the relatively dark environment of the movie theater. (That's why the filming process involves a composite gamma of approx. 1.8 from scene contrast to projection: around 0.6 gamma for the film negative in its toe-shoulder range, 1.0 for intermediate stocks, 3.0 for release stocks. The release stock serves to make contrast presentable. In stills prints, the high gamma of the photographic paper does the same: injects contrast back.)

If we had high precision displays with very high bit-depth, capable of reproducing very high dynamic range (high contrast ratio similar to scenes contrast ratio) no gamma would be required at all, be it log or whatever. Gamma encoding would be redundant as it is simply a vehicle for effective tonal quantization. Dispaying contrast, on the other hand, would still need to be taken care of, because unlike scene viewing conditions, watching conditions stay more or less the same (during an entire movie). If we had a display system which alters viewing conditions to match scene conditions, that would be redundant too.

Actually some of this you have covered here, so maybe I am being redundat and overexplaining things:

Agree, but I think those more sophisticated gammas, while suffering from the obvious flaw of being 'unviewable' without the LUT, succeed in distributing the light levels into digital word sub-ranges in more 'linear' fashion. By saying 'linear' I mean linear in relation to our eyes (and brain) transfer curve - we are not as sensitive to absolute light level differences in highlights as we are in shadows and mids, and those gammas make use of this fact.

Also, being more or less LOG-based they still allow for coding higher DR. IMO 8-bit LOG is much better than 8-bit encoded with standard gammas and that's all I meant here

Yes, that's the point of LOG. Except I disagre a bit with its applicability to 8 bits, due to the limited precision. The Higher DR (more correctly, the higher precision in blacks) is not for free and comes with limiting precision in the mids. Here is a bit more detailed text on flat curves, I've written recently, so that I don't repeat myself here . (You may skip directly to the second part "The case of flat picture styles", it is on topic).

 

[indiawilds]

Frankly speaking 15k USD seems a lot. However, nothing is free in life. One needs to use the equipment he/she has. It is true that if the 1D C would have been close to the 1D X then also some people would have panned it. Cutting out all that clutter, one can clearly see that the 1D C is a powerful package and would be shine in the right hands.

I am sure Canon will take feedback and bring down the price. If it is close to 10-11k range, then a lot of people will be more inclined to buy.

Not many people realise that the huge file size will pose its own challenge. It is not like any other DSLR that you carry a few cards and download those. Editing will be another challenge.

Shane Hurlbut has uploaded a BTS clip in his site: http://www.hurlbutvisuals.com/blog/...e-canon1dc-dslr-4k-capture-project-the-ticket
That gives a good idea what this camera is capable off and where it can be used.

 

[Emanuel]

I'm still unsure why everyone is up in arms about the price.

As far as I understand, there are (or will be) six 4K cameras, this year, under $100k (for a basic shooting setup):

Sony F65 - $85+
RED Epic - $40k+
RED One - $30k+
Canon C500 - $30k+ (?)
RED Scarlet - $16K+
Canon 1D-C - $13k+ (?)

Each camera has it's own +/- points and a price that seems to fit this relatively niche, professional market.

Why $30k+, Jay? For same reason as C300 (from $20k down to $16k in the end) and your estimation guess for 1D/C, it should end somewhere between $29,995 ;-) and $25k+

 

[vladszp]

+external recorder...
4K RAW it wont be cheap...