RED concerns

F

FatBird19

Well-known member
I'm sure that this has been expressed before, and I'm not trying to be a troll, but what if the image quality from the RED's CMOS sensor is not that great?
This camera is offering Genesis camera specs at the price of an SDX900. Doen't this make people somewhat concerned?

At that price range, I am very skeptical. Worst possible scenario, the sensor could be high grain, low color fidelity, low lattitude, bad highlight handling, and vertical smear ridden.
My hope is that the images from RED are absolutely stunning, but until I can actually see some screenshots my excitement is minimal.

It's not that I don't have faith in Jim and his team to produce the best possible camera that they are able, I just don't have faith in current CMOS technologies to produce a production quality 35mm sensor for under $18k. :D

....not trying to be a pessemist, just sayin...
 
If it wasn't that great. Jim would have pulled out a long time ago and would not invest a grip of money into sensor that was mediocre. They have tested this thing (according to reports) and we'll see some footage in the fall.

"offering Genesis camera spec" I don't think ANY camera has 2450p specs. It has 5 times the res of the sony f900.

too good to be true? seems so and I am not surprised by the skepticism.

I have faith in a guy who knows good quality when he sees it. Check out his personal page which is full of photographs and tell me if you doubt his ability and vision to create the best possible camera this world has ever seen. hehe.
 
Last edited:
Sound like pessimism to me. This team is just too good and they don't have the big corporate baggage of the other guys....they started with a clean slate. I'm certain enough that I ordered one.
Aloha
-A
 
I would keep in mind the products which Jim has brought to market in the past. He does not seem to have a history of putting out cheap imitations. Oakley, for example, is a company that sets the trends for the industry it is in.

I would also keep in mind the team that was put together for this.
You think Graeme is going to be part of a project that puts out a high grain, low color fidelity, low lattitude, bad highlight handling, vertical smear ridden image?
:)
 
Poi Boy said:
This team is just too good and they don't have the big corporate baggage of the other guys....they started with a clean slate. I'm certain enough that I ordered one.
Aloha
-A
Click to expand...

nuff said.
 
hm, alright.

What codec will they be using for the video? Maybe it's on the site already, but I don't know.
Will it have a jpeg2000, wavelet, or H.264 base? I personally would hope for jpeg2000.


And what does it record on? (again, it might say on the site but I'm not sure)
That'd be cool if they could get permission to use P2. 600mbps 4:4:4 10bit log 4k video would be awesome! :D
 
Last edited:
The codec will be variable bitrate, wavelet based.

It records to a myriad of devices, including a removable hard drive solution, removable flash drive solution, internal onboard memory, and any kind of external connection you could possibly ask for (firewire 400, firewire 800, USB 2.0, eSATA, etc.)

No P2, but there's a possibility of compact flash.

There is PLENTY of information available on this camera now, at this website, on DV Info, on HDforIndies, on Red.com, on DV.com, in downloadable podcasts, and even a parody blog. If you exhaust all of those sources and STILL have a question, well then, I'm sure plenty of people will be happy to answer it for you.

Thank you. :)
 
Haakon said:
The codec will be variable bitrate, wavelet based.
Click to expand...

Let's hope that's not true. The last thing we'd want is for this camera's potential to be squandered with lossy compression.

As far as worries about CMOS go, look at Canon's digital SLRs. They're all CMOS, and have the lowest noise out there with plenty of sensitivity.
 
stokestack said:
As far as worries about CMOS go, look at Canon's digital SLRs. They're all CMOS, and have the lowest noise out there with plenty of sensitivity.
Click to expand...

Yeah, but the cheapest Canon DSLR with a full 35mm sensor is $3000 and it only shoots 3fps.
3fps for $3000 to 60fps for $18000 is a big leap, especially considering that Canon's CMOS manufacturing is some of the best out there and is still expensive.
 
stokestack said:
Let's hope that's not true. The last thing we'd want is for this camera's potential to be squandered with lossy compression.
Click to expand...
I too have never been a fan of the variable bitrate mentality, but several of the people connected to the production of RED have said the codec is virtually lossless. Jim Jannard himself has indicated that the latest round of testing delivered "better than expected" results. I really don't think you have to worry about these guys building a 4K supercamera and then relying on a codec that is "lossy."

They have also mentioned that the codec is one of the things that will remain in constant development all the way through the preproduction cycle.
 
Based on Jim's comments that he's been satisfied with his digital still cameras but not with any of his HD cameras, and the fact that he uses a Canon dslr, I suspect that he'll be measuring his success to some extent on the degree to which RED image quality rivals the Canon dslr's.

I'm not saying we should expect it to equal the Canons, but that that may be the benchmark.
 
adaml said:
Based on Jim's comments that he's been satisfied with his digital still cameras but not with any of his HD cameras, and the fact that he uses a Canon dslr, I suspect that he'll be measuring his success to some extent on the degree to which RED image quality rivals the Canon dslr's.

I'm not saying we should expect it to equal the Canons, but that that may be the benchmark.
Click to expand...
He probably owns (will own) a hasselblad H2d 39megapixel for approx 31K (yes you heard the price correct for this top of line still camera, so perhaps someone needs to create a revolution in that market to...price-wise.)

peace

dalen
 
First, compression is a reality - get used to it. The lens compresses our 3d world of fantastic detail, as does the act of using a sensor to capture the light. "uncompressed" is never, ever truely uncompressed, other than as a step in the process where you can mathematically say no loss of data has occured. This, is nice, but often impractical.

Onto the codec - we've said it will be VBR wavelet based. This is a good thing as wavelets look good. Anyone who's used a wavelet codec can tell you that when you do push it too far, you get a bit of softness - that's it, and it holds up so much better than DCT for colour correction etc.

There's other stuff that I can't go into yet, but I don't think anyone will be unhappy with our codec decisions. As time goes on, we'll probably put up white papers on this kind of stuff, but now, we're in the thick of engineering.

Graeme
 
Graeme_Nattress said:
Onto the codec - we've said it will be VBR wavelet based. This is a good thing as wavelets look good. Anyone who's used a wavelet codec can tell you that when you do push it too far, you get a bit of softness - that's it, and it holds up so much better than DCT for colour correction etc.
Click to expand...

I don't know a lot about wavelet based codecs (ie. nothing), so was just wondering (from a post pov) how processor intensive is it and how "editable" is it?
 
Wavelets themselves are not processor hogs, and tend to be very editable. What does get tricky is various entropy encoding algorithms that compress the result of the wavelet transforms. Obviously, given that understanding, we're working on making sure that the result is editable.

Graeme
 
Yes.. Wavelets are way easier to handle then DCT.. Wavelets can also be throttled on the fly in preview to maintain a zero dropped frame relationship... kinda like when you lower preview quality in your output monitor in a NLE.

Wavelets also dont use block encoding, which sounds great but is actually a little retarded. Using block encoding opens up a range of issues that can happen with differences in joining blocks, usually in 8 pixel blocks. Wavelets sample the entire image, and create a relationship similar to a vector.. albeit a very complex vector.
 
stokestack said:
As far as worries about CMOS go, look at Canon's digital SLRs. They're all CMOS, and have the lowest noise out there with plenty of sensitivity.
Click to expand...
Hmmm, they do have a real sensitivity of ASA 100. Going up from that is just boosting gain, which increases the noise.
Btw, Canon 1DsMkII's pixel size is 7.2 µm. Mysterium's is 5.4µm. That makes the Mysterium's pixel's area to 56% of what 1DsMkII's is. If the sensors would have same sensivity and rated same, Mysterium's sensivity would be ASA 50!

Of course moving picture camera's are rated differently, because the noise isn't so noticable with moving pictures.

CineAlta is rated as ASA 400 and it has about same pixel area than Mysterium.

What if you use Scaled resolutions with RED?
Does that increase the sensivity of cmos? Like joining several pixels to be one? And the group of pixels having just one pixel's base noise?

Sooo, if you need sensivity of ASA 500, which one is noisier RED or Vision2?

Any guesses about RED's ASA ratings?
 
You must log in or register to reply here.
Back
Top