Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Message for Jarred

Status
Not open for further replies.
J

Jan Wenstad

Member
Joined
Dec 4, 2012
Messages
17
Reaction score
0
Points
0
Hey Jarred,

Recently upgraded to Dragon, which is great for the current path of tech Red is going. But, there is a better option out there.

The Foveon X3 (here is a link that you can visit http://www.foveon.com/article.php?a=70) This chip is not just a step in megapixels, it's a giant leap in image quality. With this type of technology; sharpness, color detail, and artifacts will exceed the dragon at lower resolutions. I have a long history in post production and VFX and the problems that I have ran into have never been 'Not enough resolution.' Honestly, all of the cineon files were 2k or 3k. In my personal opinion, companies have been attacking these issues by developing chips with more and more megapixels and not fixing the array pattern on the sensor. 5k is more than enough.

Anyhow... I'm not sure how successful this is in reaching you, but if it does and you would like to hear some of my other ideas, let me know.

And please redusers... remove the "fan" glasses for a moment and look at the tech and how it could really change things up for the better.
 
Jan,

How big are the files? Dragon is already a beast to manage and buy all the data. At some point, we got enough res to make really good movies. We need a more friendly codex that doesn't require super computers.

Could a new sensor reduce file size?
 
Will Keir said:
Jan,

How big are the files? Dragon is already a beast to manage and buy all the data. At some point, we got enough res to make really good movies. We need a more friendly codex that doesn't require super computers.

Could a new sensor reduce file size?
Click to expand...

Dragon by far is the easiest beast to manage.
 
Got to say I agree on the diminishing returns (increasing cost and headaches) of ever increasing resolution. Would love to see a sensor release that focused on the quality of pixels rather than number.
 
This has been around for years. And I have no doubt that Jarred and pretty much everyone else that makes, works with or sells sensors knows what Foveon is. But so far they have only been able to make small sensors their biggest now is 20.69 x 13.79 mm. And to me pixel quality is important but sensor size matters as well and the debayer problem with Cmos can be sorted with soft olpf's and oversampling... Thats what RED and pretty much everyone else is doing. Foveon might catch up but so far competition has shown that Cmos and oversampling has been the way to go and I think sensor size is a big part behind that reasoning.

Edit: also it's not really fair of them to do the comparison at the same 1:1 pixel resolution as most of todays CMOS sensors have both a more dense pixel array and far more pixels. So if they would instead compare picture quality from a common sensor size I actually thinks the good Cmos sensors of today wins big time.
 
Jan Wenstad said:
And please redusers... remove the "fan" glasses for a moment and look at the tech and how it could really change things up for the better.
Click to expand...

Better - more color resolution per total pixel count.

Worse - sensitivity, 3x file sizes for raw.

With latest Sigma's sensor you'd get around 240 MB/s for 5:1 compression, wanting to avoid going above ISO 400.
 
This forum is so interesting... It's always the same group of individuals that don't look beyond. The x3 chip that is in the SD1 has the same crop factor as 4k on your camera and Bjorn is incorrect the chip is 23.5x15.7mm. But all of this doesn't really matter, the die can be cut to specification when ordering the chips. I'm pretty sure that Red isn't in the chip manufacturing business so they must be striking deals with someone. Get into talks with Foveon, work with them to develop a FF sized chip. Eliminate mosaic patterns, color softening issues without having to solve the issue with image reduction. Increase the pixel (because the RGB is stacked on top of each other) which allows for greater light sensitivity and less grain.

For everyone with thousands of posts, stop being so narrow minded in the pursuit to create the perfect image. I know for myself, moire and grain are the two other DP's on every shoot I do. And quite frankly, I'm tiered of them lingering around the set. The worst is when Mr. Neat has to be hired for a week in post, talk about a time sucker.

All things put aside, I do realize that most of you guys are primarily shooters and haven't dug deep into image manipulation constraints. But there is a better way and embrace "change", isn't that the motto around here?
 
Jan Wenstad said:
All things put aside, I do realize that most of you guys are primarily shooters and haven't dug deep into image manipulation constraints. But there is a better way and embrace "change", isn't that the motto around here?
Click to expand...

Hey.
One of the things that some of us with thousands of posts have done, is that we have read a few. Here and around the net. And been responsible for signal chains and workflows for a generous amount of productions.

If this was the first thread on foveon or included any original/new info, I'd be thrilled. Like really.

But it does so far not, and thus... It's less interesting.

I suggest you google "Foveon+reduser" before making assumptions...

OTOH: Yup.
Conceptually it looks like a good idea, and I totally agree that CMOS is probably not the end of the road

not really sure if Foveon is the continuance, though. But I agree: It is maybe a part of the riddle.
 
Jan Wenstad said:
This forum is so interesting... It's always the same group of individuals that don't look beyond. The x3 chip that is in the SD1 has the same crop factor as 4k on your camera and Bjorn is incorrect the chip is 23.5x15.7mm. But all of this doesn't really matter, the die can be cut to specification when ordering the chips. I'm pretty sure that Red isn't in the chip manufacturing business so they must be striking deals with someone. Get into talks with Foveon, work with them to develop a FF sized chip. Eliminate mosaic patterns, color softening issues without having to solve the issue with image reduction. Increase the pixel (because the RGB is stacked on top of each other) which allows for greater light sensitivity and less grain.

For everyone with thousands of posts, stop being so narrow minded in the pursuit to create the perfect image. I know for myself, moire and grain are the two other DP's on every shoot I do. And quite frankly, I'm tiered of them lingering around the set. The worst is when Mr. Neat has to be hired for a week in post, talk about a time sucker.

All things put aside, I do realize that most of you guys are primarily shooters and haven't dug deep into image manipulation constraints. But there is a better way and embrace "change", isn't that the motto around here?
Click to expand...

Maybe I´m wrong on the dimensions of their biggest sensor, But my point was so far their sensors are quite small / far smaller than the current Cmos sensors. Also very important, they are far less dense, and less sensitive. Also as others all ready stated here, to get raw you need to save all 3 layers from that sensor... So basically a 4k image from that sensor weight as much as three 4k C-MOS frames.... Granted 4k foveon will look better than 4k Cmos, but if you take a CMOS sensor with 3 times the density, then the race is not as easily won, especially if you add the better light sensitivity of the C-MOS sensor.

And no, I don't consider my self a shooter, I run a VFX shop and have done so for over 20 years and I have had a great interest in picture quality and high resolutions for a long time. I don't know what you where doing 15 years ago, but I bought my first 4k film scanner, a Kodak Genesis it scanned 35mm film in three passes, one for each color channel and achieved it's very good results in a very similar manner as how the Foveon sensor work. Also got a little foveon pocket camera when they first came out so I think I have a pretty good idea of how they work.

And if you think of it. If they actually where much better than Cmos... don't you think the tech would be implemented in pretty much every new high performance camera out there?

And I dont know if it's true but both Red and Arri says that they make their own sensors.
 
Jan Wenstad said:
This forum is so interesting... It's always the same group of individuals that don't look beyond. The x3 chip that is in the SD1 has the same crop factor as 4k on your camera and Bjorn is incorrect the chip is 23.5x15.7mm. But all of this doesn't really matter, the die can be cut to specification when ordering the chips. I'm pretty sure that Red isn't in the chip manufacturing business so they must be striking deals with someone. Get into talks with Foveon, work with them to develop a FF sized chip. Eliminate mosaic patterns, color softening issues without having to solve the issue with image reduction. Increase the pixel (because the RGB is stacked on top of each other) which allows for greater light sensitivity and less grain.

For everyone with thousands of posts, stop being so narrow minded in the pursuit to create the perfect image. I know for myself, moire and grain are the two other DP's on every shoot I do. And quite frankly, I'm tiered of them lingering around the set. The worst is when Mr. Neat has to be hired for a week in post, talk about a time sucker.

All things put aside, I do realize that most of you guys are primarily shooters and haven't dug deep into image manipulation constraints. But there is a better way and embrace "change", isn't that the motto around here?
Click to expand...

No offense taken at all. And I'll even go that extra step further and say it's nice people are making suggestions like this from time to time.

However, in your first paragraph there's a few "wrong statements" in how Red goes about making it's sensors these days, especially when it comes to R&D and development. I'll leave that there. However, there are a few handfuls of people working nonstop to develop bleeding edge technology behind the scenes to squeeze out the highest quality color, dynamic range, and picture quality that they can.

There's also some of us with "thousands of posts" who have been involved with building a couple of cameras and also have been around a few prototypes of cameras from various manufacturers some of which never even made it to market due to chaos.

In some ways the Foveon tech represents some very logical steps, however in manufacturing, patents, technological capability, and various other motion based oddities the juice isn't exactly worth the squeeze.

Specifically when it comes to the Foveon tech itself, while pixel sharpness, general potential color saturation, and reproduction are good there are of course issues. And that comes in the form of "other shooting conditions", issues you don't read on the sensor tech "sell sheet". The sensor itself has rather poor dynamic range and super saturates (gets rather confused) on tricky colors commonly found in highlights and lower lighting conditions. Meaning near middle grey adequately exposed is going to be fine, but the further away you get into some hairy business.

Here's an example from the Sigma DP3, which is using the latest Foveon tech (Courtesy Eugene Fratkin's review):


And from a Fuji X100s for comparison:



That right there is a "no go zone" actually when it comes to the demands needed by serious digital cinema cameras.

That said there are developments on this front using similar concepts. Canon's more recent 5 layer design is going to be interesting as it's being designed to do a few things outside of what you'd expect from any camera on the market today as it taps into also capturing and manipulating IR and UV light. That patent was publicaly released in 2014 and it's not deployed in a camera system yet, but it's interesting.

Nonetheless, you can be rest assured that there are people at Red who literally only deal with sensor and future sensor design.
 
62 is not more than 80. ;)
 
Wow, Foveon, haven't heard that name in almost 15 years. I was shooting a megavision digital back at the time when they came on the scene looking very promising. Seems like they haven't progressed very far since then but who knows.
Aloha
A
 
That is one hell of an image, it's hard to believe that came straight out of the camera. Makes me want to try and recreate that garbage. Anyhow, everyone here is seriously missing the point. Bringing in a subject matter into the forums that doesn't highlight how AMAZING Red is... is like walking into a forest with 10lbs of raw meat strapped to your belt, you never make it 100 yards before you're attacked by a bunch of wolves.

First off: fill your bellies, you might be less aggressive.
Second: the idea behind the tech is a pretty sound solution even though the tech has some issues, but that is exactly what Red is good at. Solving issues!
Third: we all know what happens when you cram megapixels in a sensor. We might end up with a cell phone camera, which is great if you do youtube cat videos.
Fourth: I like the chatter about how they remember hearing about the tech 15 years ago, but it is also worth noting that the Bayer system was developed in the 70's (45 years ago, yeah, 45 years ago) Megapixels does not always mean better image, that is a prosumer marketing war and I would rather not see Red follow suit.

I would love to see Red's inventive, problem solving, no nonsense attitude crack the nut that Sigma is slowly gaining ground on. Grab the bull by the horns and knock the socks off motion image processing. FFS people, have faith! Logic only tells me that the race towards megapixels will only result in diminishing returns in both image quality and computation power in resolving the image.

Oh, forgot to mention... for the love of god, FF sensor please... I can't express how annoying it is that FOV, DOF and Depth Perception never feels correct. A 50mm should still be a 50mm, if needed, develop a scaling algorithm and add another chip to the camera, whatever it takes.

Not sure if this will strike a cord with you guys, but all of this negativity is neither helpful for Red or yourselves.
 
Status
Not open for further replies.
Back
Top