Welcome to our community

Be a part of something great, join today!

  • 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 :)

Big Change...

What is lost

What is lost

Sometimes I get the feeling some people don't get it at all. Invert your statement and you get closer to reality. Canon didn't market their sensation- they released it! They didn't use a fraction of the size of their FULL FORMAT sensor- they put it to use the best way they know how to, at the end of 2008, within certain parameters of price and so on.
That is exactly what Jannard and are saying. While a technical discussion is not warranted, this is 'cheap' or 'flaky' from an engineering standpoint. Taking a full or half frame resolution and scaling it to 1080p would be tolerable.

Canon is giving this customer EXACTLY what he wants. And judging by the reaction, I'm not alone....I'm buying a camera for $3500 that I can use for lots of stuff. Where do you think my 'movies' will end up? "In a theatre near you"?? I'll export to 720p and distribute my movies online via Vimeo or whatever for anyone to see/comment. I will showcase higher bitrate files on my laptop when I'm visiting friends and family.
Lack of professionalism is never an excuse for a company cheating you of quality. Particularly with Canon.

It's the FORMAT (full frame SIZE) I want, not absolute quality. Of course I want that too, but I know I can't afford it...
I'll have what I need with Canon in two weeks. Taste that: 2 weeks.
I have news for you though, based on my understanding of focus and Point of Confusion, taking a much smaller 'dot' or photosite spanned over a large sensor and scaling+condensing (likely a delicate mathematical balance) those dots is going to have some effect on the DOF you get.

You can get you cake now, from the mass-produced market and it will be chocolate with frosting and good enough. Or you can wait a while for a group that really cares what you want to give you richer chocolate, fresher taste, the shape and consistency you want, and in all a much nearer to perfect experience. And such is the difference between Canon/Nikon/Sony/etc and RED. Jim is a connoisseur of images, and he wants to cater to other connoisseurs, most of which are professional cinema makers. I don't care what you buy, its your money, and that's the glamor of freedom. But I want people to discern 'cool' and available from excellence.
 
Apple goes back to the drawing board all the time.
I haven't seen this. Apple usually picks a direction and sticks with it, and either it was a bad idea and they late in the game switch (Intel processors) and make it sound like their the saviors of computing or it take off and they refine it. The first iPod was huge (by comparison) and slow. But they refined it it a little and it finally took off.[/QUOTE]

What's the upside? You get to go about minding your own business, till one day there's announcement from right out of the blue: "We've spent the last three years building this thing. It ships today."
A) RED does this. We went from thinking Scarlet and EPIC were a 'prosumery' 3k and EPIC a RED Two. Then BAM! we have a large & small brain in a modular system, and I'll bet that took quite a while to develop. And further, they release revisions at a surprise.
 
After reading the statement from Jim about Canons "parlor trick" (skipping pixels to get 1080p recording from their "full" sensor area)... I've started thinking...

Actually... Even though it isn't exactly in the RED-way. Why not use that same parlor-trick and better Canons result. The Canon sollution does have it's merits, though it's hampered by lack of controll of exposure (as I have been told at least. I don't know if it can be worked around like with the semi-auto HV20), crippling compression, limited FPS control and no raw.

There has been much talk about a fictional s35 3K scarlet because of the percieved features it would bring with it. Even as Brainburst continually puts forward the 4/3 way of doing things. A s35 scarlet in 3k has the main points of lower storage-needs (than the promised 5K at least) and ability to use s35 optics and therefore DOF (which I tend to feel is overrated IMO).

So, what I'm getting at is that the red cameras has always been able to do "windowing" to get higher FPS. And in all fairness, In my mind, "windowing" is also a parlor-trick. You trade off resolution and FOV for FPS. But it works. So... What's stopping The Team from putting pixel-skipping into the camera as a firmware upgrade feature? That would give, let's say, the 6K scarlet that has a maximum of 30 FPS, a mode of recording "pixelskipped" 3K in somewhere of at least 60 FPS (and assuming that 3K being quarter of 6K it maybe would be able to boost FPS up to, say, 30*4=120fps, not counting the effect of bayer-grids of course). utilizing the full FOV of the Full Frame sensor. In RAW, with full exposure control.

True, the dynamic range and low light capability isn't boosted at all with this method. But at least everyone that craves the "nothing in focus"-look would be happy. And we all get full-frame slow-mo. Without the over-stated rolling shutter issue of the Canon;) I don't know if the voodoomagic RedCode would be thrown into a fit though, having to skip pixels to reduce resolution instead of focusing of a smaller area of the sensor...

And maybe it would lower, what I imagine must be a concentration of energy on a small space on the sensor and therefore concentration of heat. Instead spreading out the energy and lower the overall temperature of the sensor.

This would only entail a firmware update I think. Not limited to the 6K camera. A 2/3 3K scarlet could in my mind pixel-skip to 1.5K RAW and 240 fps. 360 burst. And the Sillily large 617 Epic could get 14K feeds of 50fps. All utilizing the full sensor area.

That is... if I've understood the tech-side of this feature in good enough way. I'm sure that it has all been discussed by the RED build team at great lengths and for some reason or another (I imagine it could totally screw up the lovely redcode magic for instance) the windowing was chosen. I just hope that the "parlor trick" isn't scrapped alltogether though if it actually would work as I have outlined above. Just because it was the way that Canon did it, and we'll relegate it to the pile of don'ts right next to "pixelshifting"...

just my 13 ören on the matter:)
 
What's the upside? You get to go about minding your own business, till one day there's announcement from right out of the blue: "We've spent the last three years building this thing. It ships today."
A) RED does this.

With all due respect, RED clearly doesn't do this, evidenced by the fact that we're being told about new products whose specifications will likely (read: certainly) be changed significantly before these products reach the end user, or are even available to purchase. SCARLET and EPIC are still very much in development; the decision and subsequent announcement about their potential form factor doesn't change this.

I don't have a problem with this, as I think it's responsible to keep existing customers and prospective customers informed about upcoming changes to a company's product lineup (which Apple unfortunately doesn't do).

However, I don't see any need to announce the announcements (ie. "stay tuned for an announcement on the..."). This just seems to produce unnecessary wild speculation that tends to lead to arguments.
 
Maybe its not a new camera or hardware related at all. Maybe a rollout from Adobe, or maybe Jim has struck an amazing deal with AVID that could substantially benefit all RED/SCARLET/EPIC users or maybe some other MAJOR new industry agreement. Maybe RED has received news of a technical Oscar nomination (would be fun for Jim)? Who knows, but whatever it is it definitely concerns the very big picture, I.e. major industry news, not a partial design re-think.

However, a REDCODE chip/board (as some one has also suggested) would also be huge…
 
After reading the statement from Jim about Canons "parlor trick" (skipping pixels to get 1080p recording from their "full" sensor area)... I've started thinking...

Actually... Even though it isn't exactly in the RED-way. Why not use that same parlor-trick and better Canons result. The Canon sollution does have it's merits, though it's hampered by lack of controll of exposure (as I have been told at least. I don't know if it can be worked around like with the semi-auto HV20), crippling compression, limited FPS control and no raw.

There has been much talk about a fictional s35 3K scarlet because of the percieved features it would bring with it. Even as Brainburst continually puts forward the 4/3 way of doing things. A s35 scarlet in 3k has the main points of lower storage-needs (than the promised 5K at least) and ability to use s35 optics and therefore DOF (which I tend to feel is overrated IMO).

So, what I'm getting at is that the red cameras has always been able to do "windowing" to get higher FPS. And in all fairness, In my mind, "windowing" is also a parlor-trick. You trade off resolution and FOV for FPS. But it works. So... What's stopping The Team from putting pixel-skipping into the camera as a firmware upgrade feature? That would give, let's say, the 6K scarlet that has a maximum of 30 FPS, a mode of recording "pixelskipped" 3K in somewhere of at least 60 FPS (and assuming that 3K being quarter of 6K it maybe would be able to boost FPS up to, say, 30*4=120fps, not counting the effect of bayer-grids of course). utilizing the full FOV of the Full Frame sensor. In RAW, with full exposure control.

True, the dynamic range and low light capability isn't boosted at all with this method. But at least everyone that craves the "nothing in focus"-look would be happy. And we all get full-frame slow-mo. Without the over-stated rolling shutter issue of the Canon;) I don't know if the voodoomagic RedCode would be thrown into a fit though, having to skip pixels to reduce resolution instead of focusing of a smaller area of the sensor...

And maybe it would lower, what I imagine must be a concentration of energy on a small space on the sensor and therefore concentration of heat. Instead spreading out the energy and lower the overall temperature of the sensor.

This would only entail a firmware update I think. Not limited to the 6K camera. A 2/3 3K scarlet could in my mind pixel-skip to 1.5K RAW and 240 fps. 360 burst. And the Sillily large 617 Epic could get 14K feeds of 50fps. All utilizing the full sensor area.

That is... if I've understood the tech-side of this feature in good enough way. I'm sure that it has all been discussed by the RED build team at great lengths and for some reason or another (I imagine it could totally screw up the lovely redcode magic for instance) the windowing was chosen. I just hope that the "parlor trick" isn't scrapped alltogether though if it actually would work as I have outlined above. Just because it was the way that Canon did it, and we'll relegate it to the pile of don'ts right next to "pixelshifting"...

just my 13 ören on the matter:)

The problem with line skipping on the full frame sensor ala Canon is that you end up with bad aliasing artifacts, like grabbing a single field of an interlaced image. Aliasing and pixel blocking get smoothed over in the 1/4HD compressed clips posted on the web, but they are very evident in the frame structure of the 1080 originals.
 
There have been some "1080p" movies from the Canon posted that show severe aliasing, and look quite nasty. http://s3.amazonaws.com/movies.dpreview.com/CanonEOS5DMarkII/Video2.MOV is about the best example for that - the multi-coloured moire on the moving train looks like an over-exposure zebra indicator got recorded, it's that bad. The moire on the shuttered door is just ugly. That's exactly what you get if you skip lines on reading out, and the funky colours are what you get when you skip in strange patterns due to it being a Bayer sensor.

Now, let's take this one step further. If you skip lines on readout of a CMOS, although you've not changed the read/reset time, the skew is visibly less (less lines read = less visible skew). We've all seen the movies of the jello and skew on the Canon, but it's not "too bad" so we think that it has a fast read / reset time. Obviously, it does not have a good read reset time as they can't match the performance of the RED, even when skipping the lines to make the skew appear less. Now, you look at the Nikon D90 with it's quite ghastly skew. Ouch indeed. And they look to be line skipping on the readout too with some quite hideous scaling artifacts to show for it. I wonder how bad their read / reset time is....

Graeme
 
Graeme, are you getting a 5D2?

Do you think they are skipping pixels to make this video mode work?
 
Nope, not getting one. I have a 1D MK III for photos for now, and a RED for motion. Until Scarlet and Epic, that's a good enough combination for me for now.

Are they skipping pixels? Well, only Canon knows for sure, but the evidence points towards them either skipping pixels, or using the world's worst scaling algorithm know to man. Either way, it's less than optimum. I also severely doubt you'll measure a proper 1080p from the movies - they have the look and feel of a lot lower resolution to me. They certainly look nothing like the 1080p you can derive from a RED 4k output properly downsampled.

Graeme
 
There have been some "1080p" movies from the Canon posted that show severe aliasing, and look quite nasty. http://s3.amazonaws.com/movies.dpreview.com/CanonEOS5DMarkII/Video2.MOV is about the best example for that - the multi-coloured moire on the moving train looks like an over-exposure zebra indicator got recorded, it's that bad. The moire on the shuttered door is just ugly. That's exactly what you get if you skip lines on reading out, and the funky colours are what you get when you skip in strange patterns due to it being a Bayer sensor.
Graeme

Sweet!! I love that look.

So are you guys planning a "line-skip" build to compete with these guys:bleh:

I.Bloom
 
Sure, there'll be a big switch on the front labeled "crap mode". :-)
 
They certainly look nothing like the 1080p you can derive from a RED 4k output properly downsampled.

Graeme

Graeme -

Sooooo... how do you properly downsample your Red footy? Just wanted to hear from the man. :calm:
 
Well, the "half high" mode does a nice downsample direct from the 4k to 2k. Beyond that, I'd pick your filter from the options available in RedAlert and see which looks best on your particular footage. Any of them will look better than the link I posted above....

Graeme
 
Tell you what, guys, how about we not stoop to the level of others by putting down the competition.

Instead, let's focus on being the best, and let that act speak for itself.

I don't like seeing RED, or those on it's payroll turning this place into C.com.

We're better than that.

Jay
 
I picked up a 5D2 a couple days ago.. Video quality aside.. which isn't horrible.. but not to what we are used to around here.. Biggest issue though is that it's not really "setup" to shoot video.. You can't just set your shutter and ISO and go.. its pretty much auto-everything exposure wise. There is an exposure lock, but it is pretty tricky. Granted.. these are software decisions that have been made, but regardless, I have a feeling people that might be buying the 5D2 for Semi-Pro Video mostly might be a little disappointed.. just in terms of actually using it. the Video mode is pretty much the same as in Canon's digital ELPH line of cameras, but with a big sensor.

Now for Still images.. the 5D MII is a pretty impressive camera. The 1Ds is going up on Ebay.
 
Back
Top