Reasons to go for C300 over RED Scarlet X?

[Meryem Ersoz]

I guess that I missed the moanfest...so I'm lacking some backstory here.

Said more than my share...and will retire my torch....don't really like playing RED defender, but hope that I added a little bit of additional information to the heap, for those of you trying to weigh options. It is a lot of money to spend, on either system, and the decisions are getting tougher to make.

[Dylan Couper]

Originally Posted by David Heath

Arguably a bit of both - yes, an explosion of people shooting large format video, but possibly more people in total as well. The ground glass adaptors involved not just their own cost, but that of the base camera as well - with DSLR video that was a huge drop in cost, and it's quite likely led to users coming along who previously hadn't been able to afford the equipment.

If I remember correctly, about 4 years ago, before the 5D2 came out, my Brevis cost about $1200 and the smokin hot Canon HV20 was about $800. Not exactly unaffordable compared to a $1600 7D if you wanted to shoot Super35. I can't join you in the theory that all of a sudden, tens of thousands of people woke up one day and said "Hey I can shoot Super35 on this whack form factor DSLR for under 2 grand... I think I'll become a filmmaker!"

I will say that it has migrated a lot of still photographers into the video world, since they now own tools that cover both... and at the same time has moved a lot of videographers (less so) into stills.

[Alister Chapman]

Originally Posted by Don Miller

Assuming the C300 is clearly more filmic than the F3, it should sell well. Specs are secondary. Canon competing in this way is historically exactly their thing. 5dII video happened because the images were beautiful, despite the many flaws.

Why do you assume the C300 clearly more filmic than the F3? Some of the footage I've seen is much more video like than the F3, clipped highlights is one video trait that many are seeing from the C300. I think we need to wait before we judge image quality until more units are actually out in end users hands.

5DMkII images can be beautiful in the right conditions, but all it's issues have limited it use to a narrow range of applications and for many mean it's a non starter. It created a niche in the market that was not really there before.

Specs are not secondary. Good reliable specs will tell you a lot about how a camera will actually perform in your hands as opposed to looking at videos on the web created by experts with big budgets who can tailor the shoot to get the most from the camera, avoiding areas that may cause issues. Take the 5D, independent resolution tests tell us that the resolution falls short of what most would expect from a 1080P camera. Yet on the web this is difficult to tell. If your producing web videos then that's fine, but present on a big a screen and the difference becomes much clearer. Of course one of the issues is getting reliable and meaningful specifications as opposed to the carefully selected and sometimes ambiguous manufacturer specs.

[James Millward]

As a complete outsider looking in ( a novice, with not a chance of owning either camera), the choice seems simpler than people are making out.

If you want a camera that will work 'out of the box', that will be much simpler to use, and will have a faster work flow without major $ investment in raw workflow then the c300 is the better choice.

Run and gun? wedding/event videography? etc then I can not imagine for one second using any red camera. the reliability issue alone woryy the hell out of me!

If you have more time to invest per shoot, have repeatable conditions, need higher res and absolutley must have raw, then of course the red is for you.

The notion that either of these camera somehow far worse, or far better images than the other seems strange to me. You could make beautiful images with either.

James

[Chris Hurd]

Originally Posted by Dom Stevenson

Further, AFAIK nobody here has actually seen a single frame of footage from this camera we've been hearing about for several years, and we shall be none the wiser till we do.

Actually there's a lot of sample material readily available. Scarlet output = Epic output, and there are plenty of Epic clips online coming from a wide variety of sources. The only differences between Epic and Scarlet are frame rate, 5k video, window size vs. res, etc., but the image "flavor" and quality of the two cameras are identical.

[Don Miller]

Originally Posted by Alister Chapman

Why do you assume the C300 clearly more filmic than the F3? Some of the footage I've seen is much more video like than the F3, clipped highlights is one video trait that many are seeing from the C300. I think we need to wait before we judge image quality until more units are actually out in end users hands.
................

I don't assume, I predict :)

The C300 sensor makes many more measurements compared to the F3. Also, as I've said Canon is telling us it's noticeably better by their pricing. The only people who have compared the F3 and the C300 are Canon. Either the C300 has better IQ, or the Canon people are foolish.

The trouble with specs, especially Red specs, is they don't illuminate the trade offs. What price does the Red design pay in basic IQ to do 120 fps? In power consumption? In low light capability? What's the upside of the C300's lowly specs? Are they being cheap, or making sure the 99.9% of shooting is the best it can be?

I expect the clipped highlights is from pre production cameras. It's unlikely in 2011 that Canon would build a system that has a fundamental problem with gently clipping highlights. But until the F3 and the C300 are shot side-by-side we don't know. It does seems the images from each will be distinct, which is nice.

I think in a few years debayered sensors will be old tech. Measuring just one color at each pixel will soon be better understood as a compromise. This may be a problem for Red, as I can't see them being ready to handle a 32mp quad 1080p sensor.

I'll keep pointing out that there's likely more information off the C300 sensor (pre compression) than the 4K Scarlet. And way more than the F3. Debayering is building out a bigger file, but not adding any information, just data estimates.

[Dom Stevenson]

Good point Chris. So the images will be identical? even so, friends of mine who've been renting Reds have done so primarily due to the frame rate (slowmo) options which the Scarlet doesn't do as far as i'm aware?

James Millward

Yes, both will be wonderful cameras, and like you i can't currently justify buying either, however i'd take the Canon over the Scarlet any day. If i needed a Red i'd rent one, as everyone i know who needs one does (even though they can easliy afford to buy several). and i'd rent one with the resolution/frame rate options. This model seems to fall between two stools for me.

I'm guessing in 18 months time there will be a lot more people kicking themselves for buying a Scarlet than there will be regretting buying a C300. The Sony F3 is a formidable camera too, so it would be a Sony or Canon decision for me. Having said that, i expect there will be amazing things done with the Scarlet by a relatively small number of people, and many others buying it without really getting anything from it they couldn't have got from Sony or Canon with a lot less hassle. Just my 2 cents.

[Don Miller]

Originally Posted by Dom Stevenson

..............

I'm guessing in 18 months time there will be a lot more people kicking themselves for buying a Scarlet than there will be regretting buying a C300..........

It's surprising how many people were willing to put up with ground glass adapters. I'm amazed at the number of people shooting weddings with DSLR. I think if new Scarlet users are realistic about data/power/computer needs before purchase they will be happy. My only concern for for people who can't really afford a Scarlet, but plan to rent it out. Red One published rates don't reflect reality. Scarlet rental prices will likely follow a similar price curve.
For the corporate buyer/producer, the question is if the more expensive ongoing cost of the Scarlet makes more money than the F3/C300.

[Dom Stevenson]

Don

But why would anyone rent a Scarlet?

There are higher end models that will only be marginally more expensive to rent and they have advantages over the Scarlet (like super high frame rates). If i didn't need those features i probably wouldn't rent a Red at all. That's what i mean about the camera falling between two stools. Rentalwise it's likely to be overpriced for what it does, So those with low budgets are likely to either go for cheaper options, while those with the cash will rent the higher spec Reds. There will also be loads on the market due to the - relatively - low cost price, so i doubt this will be much of a success for people planning to buy for rental.

[Alister Chapman]

Don, we will have to wait and see how the C300 sensor performs in the wild. But one thing to consider is that the Bayer process was in part developed to overcome issues created by the use of a 2x 2 sampled CFA (colour filter array) as used in the C300. Bayer has the big advantage in that it can help compensate for the leakage and cross colour contamination caused by the imperfect colour filters in a CFA. This compensation can dramatically improve colour fidelity. In addition assuming Canon are grouping the RGGB pixels under a single micro lens then there will be the issue of double green sensitivity compared to R and B. This may have an impact on noise as you must either reduce the green sensitivity or increase the R and B gain which will increase noise.There are pro's and cons to 2x2 CFA and Bayer. 2x2 CFA has been around since the 90's yet is rarely used. A really good example of the issues that can be caused when not using bayer is Sony's F35 which has 2 pixels for each colour in a stripe array, yet has some pretty bad aliasing artefacts.

Highlight handling will be a function of many factors including the sensor itself and the way the signal is processed and the gamma curve used. The blown out highlights could be many things, simply bad lighting, bad post production or web encoding. But it could also be poor signal processing in the camera. The whole lack of a 10 bit output does ring alarm bells as to what bit depth the DSP is working at.

Ultimately we'll have to wait and see, but one thing is sure sure, with only an 8 bit output your options are more limited than with cameras with a 10 bit output. However if you don't want or need to record externally the C300 ticks many boxes.

[Buba Kastorski]

Originally Posted by Chris Hurd

The only differences between Epic and Scarlet are frame rate, 5k video, window size vs. res, etc., but the image "flavor" and quality of the two cameras are identical.

Well here you go, that's the reason #1 to buy "Epic" picture for a quarter of a price, to own a camera that huge Hollywood blockbusters were shot with, even though most likely I will never shoot nothing larger than "Friendly lu lu spa " commercial, or "Giuseppe and Maria Wedding Highlights" , but just to know that I have that tool, and I might shoot something better than that is great, AND the picture is amazing,
(please don't tell me about the last Sundance and that 7D movie - i know)
am i buying a dream - maybe, but I promise as soon as C300 is available from any rental house I will put it side by side with Scarlet X and will shoot "donkey balls" out of both of them , and 1D, and 5D, and EX1 looking for reasons to go for C300,
unless of course someone else will do that before me :)

[David Heath]

Originally Posted by Alister Chapman

But one thing to consider is that the Bayer process was in part developed to overcome issues created by the use of a 2x 2 sampled CFA (colour filter array) as used in the C300. Bayer has the big advantage in that it can help compensate for the leakage and cross colour contamination caused by the imperfect colour filters in a CFA.........

As always, it rarely comes down to a single factor, and same with Bayer versus Direct Read (as here). It's impossible to say "x is better than y" - and same here.

Additionally to Alisters comments, then it's also worth pointing out that for given sensor dimensions (let's assume 100x100) then true deBayering will squeeze a higher luminance resolution out. To a first approximation, typically about 80x80 in that case for luminance, 50x50 for chrominance. Hence the F3 sensor dimensions - such that 80% gives roughly 1920x1080. The C300 approach will give 50x50 for luminance and chrominance.

So does that mean I'm saying the C300 sensor arrangement is worse? Well no, and this is where it starts to get complicated.

Direct Read is far simpler to implement. And that means simpler (hence cheaper) electronics, and crucially lower power consumption. The key is in coupling it with a sensor of optimum dimensions - which the C300 does.

What I also foresee is a 2nd generation camera which ADDITIONALLY will deBayer the 4k sensor (or record it RAW) to give a 4k output. (OK, it'll be only about 80% of the res, but that's what most people refer to as 4k). Or just switch it to 1080 Direct Read when that's most appropiate. The F3 sensor is less versatile - it has to be deBayered, and is then 1080, end of story.

Originally Posted by Alister Chapman

....2x2 CFA has been around since the 90's yet is rarely used.

I think you may be surprised how much it's being used, but not mentioned. It's one of the better ways of deriving video from a still sensor with a high pixel count, and is exactly how I was told nearly a year ago that the AF100 does it. The difference from the C300 in these cases is that alternate 2x2 blocks are omitted horizontally and vertically, so only 1 block in four actually gets read. The sensor dimensions for the AF100 seem to be 4700x2644 (for 16:9 cropping), so 2350x1322 2x2 blocks. Read out on an alternate basis and that gives you 1275x661 of blocks (and hence true resolution) - pretty well exactly what has been measured. It's a decent way of simply getting video from a designed for stills sensor, but can't compete with a sensor specifically made for video.

The AF101 is the one I have the figures for, but I understand it's far from the only case of the principle. It's an acknowledged technique for video from stills cameras. (And far better than earlier techniques that skipped whole lines asymmetrically.)

[David Knaggs]

Originally Posted by Chris Hurd

... Scarlet output = Epic output,
... but the image "flavor" and quality of the two cameras are identical.

The thing that causes my reservations on this point were posts where Red One MX owners had just received their early Epic-Ms and were commenting on the improved images of the Epics, even though both cameras had MX sensors. The consensus eventually put it down to the improved electronics and signal processing of the Epic.

The fact that they are reducing a lot of the electronics and boards for the Scarlet (not to mention using sensors which didn't make the specs for the Epic) is what is currently giving me some doubts and reservations regarding the image quality and flavor of Scarlet and Epic being an exact match.

As I mentioned in an earlier post, I'm a big fan of impromptu-style shooting (by someone who really knows how to set up the camera, expose properly, etc.) with minimal or no grading as being a more genuine demonstration of the camera itself. The clip below is a perfect example. It's from an Epic, where an excellent shooter was standing on the beach and was waiting for 5 minutes for the rest of the crew to show up. So he took a few minutes of impromptu footage in the meantime. Admittedly, it's slow-motion (overcranked) - which the Scarlet can't do - but if someone can post Scarlet footage which matches the flavour and quality of the actual images in this clip, I'll be 100% sold on Chris's statement. And I sincerely hope that I will be!

RED EPIC and a spare 5 minutes on Vimeo

Regarding the C300, I'm most excited by the fact that it's a 4K sensor oversampling to 1080p. I'm a big fan of oversampling and I reckon that the C300 is likely to make sensational-looking images for this reason alone. Most posters in this thread don't seem to have noticed or mentioned this fact. My initial interest in the Red One all those years ago was the fact that you could take 4K images and oversample them into a 1080p output.

Well, the C300 does this without the massive files, extra computing power and extra computing time. It's a simple 50Mbps 4:2:2 onto CF cards, a battery which (per Jim Martin) runs for 7 hours and costs $150. The recent price drop of the C300 (apparently?) makes an even more compelling case.

However, the final image quality and "flavor" of the C300 won't be determined by oversampling alone. How good are its electronics and how good are the gamma controls and what sort of scene files or Picture Profiles are able to be constructed? Not to mention Alister's concern that the C300's sensor array might be prone to artifacting.

That's why I hope that Canon release a few C300s "into the wild" sooner rather than later. So that we can see what some good shooters can do with this camera. Without the colorists.

[Don Miller]

I don't see how the C300 is oversampling, even if the term is just an analogy.

[Tim Le]

Originally Posted by Meryem Ersoz

1) Tim, you don't need a side handle or module to power a brick. That is not accurate.

That is true. I should have been more precise in my wording. I was referring to the internal batteries and the RedVolts do require the side handle or a module. My point is the C300 has an internal battery bay and Scarlet/Epic does not. Therefore, the C300 has an additional size advantage over Scarlet when you consider this internal battery bay, which can power the C300 for 3 hours. In the behind-the-scenes footage of the C300 films, I noticed they used the internal battery very often. But this could be due to the camera needing 8.4V in and they didn't want to deal with stepping down the higher voltage bricks.

Originally Posted by Alister Chapman

There are pro's and cons to 2x2 CFA and Bayer. 2x2 CFA has been around since the 90's yet is rarely used. A really good example of the issues that can be caused when not using bayer is Sony's F35 which has 2 pixels for each colour in a stripe array, yet has some pretty bad aliasing artefacts.

The C300's green photosites are offset half a photosite vertically and horizontally, which "cancels" out aliasing for that channel, according to Larry Thorpe. Does anyone know if other 2x2 CFA non-debayering sensors do that? I keep wondering what are the downsides to Canon's sensor design because it seems so simple and elegant.

Originally Posted by Alister Chapman

The whole lack of a 10 bit output does ring alarm bells as to what bit depth the DSP is working at.

Canon says the DSP is working at 12 bits for the red and green and 13 bits for the green. All the non-linear processing happens at that bit depth.