By the way, I tested out the video on all the new compact large sensor cameras at PMA. The short of it all? Puke. Yep. I guess I’ll be holding my EVIL camera money in my pocket a little longer. Canon? Canon?

Right?

–Zeke

I guess 144 petabytes of storage on one card should cut it.

What a fun series of events…

Zeiss Compact primes available in CF mount
+
CF Card capacity goes from 32gb to “Ludicrous”
+
Photokina coming up
=
1Ds mk IV video mode is gonna be awesome.

Thoughts? Might the 1d mk4 and the 7d (both with I think two image processors right?) more apt to faster video (or larger/raw) than they are right now?

Crazy stuff…

Red Raw does have compression in it. It is my understanding that raw sensor data is made up of visual information and non visual information. It is my opinion that the Red Raw codec is efficient because they dump the non visual sensor information retaining the raw visual info.

This is interesting because Red Raw is 12bit, smaller and has a faster data rate than 10 bit uncompressed files.

Canon can learn from Red. They don’t need raw per say. A slightly compressed codec would suffice for now. This brings me to my next point.

I’m a freelance colorist and there is huge interest in using the 5DmII for commercial work here. I’ve already worked on a number commercial jobs shot on the 5DMII. The only complaint is that the footage is severely compressed (h264). If Canon found a way to reduce the compression this camera would be heavily used for commercial work.

I too did a little investigating last night on this issue.

@Mark has already pointed out the Red writes to CF cards, tho I didn’t find out exactly how much time they get per card. I’ll go with your estimate for now.

Yes the Red sensor is smaller than the 5D2, but I think we’re not worried really about size here… the 5D2 sensor can record a still image up to 5600×3700 (roughly)… and a 1080 HD image is only 1920 x 1080 – so the canon is NOT currently writing out and compressing the entire image size… it is throwing away a bunch of data to get the frame size down to 1080.

According to wikipedia, the red one is writing out “2k” images at the smallest resolution that are: 2K 16:9 (2048×1152) — which is just slightly larger than the HDTV resolution of 1920×1080… so the small resolution on the Red is almost the same as the current 5D2 and it can write TVHD sized video at 12 min / 4gb (aprox).

And the highest resolution on the Red One is 4K 16:9 (4096×2304) – which is still smaller than the full size of the Canon 5D2 – but we all know that the canon isn’t currently writing out video in that size. And Canon is compressing with H.264

So, the question really is how much more data is there in RAW video as compared to H.264 that Canon is currently writing out isn’t it?

Per RED’s FAQ (here), their sensor is APS-C at 24.4mm x 13.7mm. That’s just a bit more than half of the sensor real estate on a 5DM2.

Can I interpolate from these stats that a RED sensor the size of the sensor in the 5DM2 will spit out about 4gb per minute and thereby fill an 8gb CF in 2-2.5 minutes? I don’t know.

“Calling all REDRAW experts.” Feel free to jump in here and set me straight.

Seems to me that while a large CF might be an option, as it in with the Red, I’ll continue to think that an external data recorder will be a must-have for projects of any size. Of course, there’s also the possibility that by the time this phantom camera becomes a market reality that Moore’s Law will have dropped the price of 64gb and 128gb CFs down to the everyday-affordable range.

One thing to note, RAW is very different kettle of fish from 10-bit uncompressed. If you look at the relative file sizes for Red RAW, Cineform Raw or any of the other formats, they are much smaller than 10-bit uncompressed. Just like the DNG from a camera is smaller than a similar quality JPEG (the operative word is similar).

To back this up. A lot of Red One owners shoot on CF cards in the size ranges you were talking about, so the Red RAW HAS to be smaller than uncompressed.