Welcome to our community

Be a part of something great, join today!

Register 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 :)

Why couldn't somebody take a 7D/5D/GH1/D3s/etc., remove the sensor and build a camera

The other unanswered question is do you want to shoot movies or spend months or years trying to compete with some of the best engineers on the planet who have serious resources at their disposal?
 
Just to add - it's not just the fps, but the sustained fps. In raw or jpeg these cameras have 10-13fps for burst. They're still way short of the 24fps minimum hence the line skipping we see in the zone plates. Not only does this fake 3 times the fps, but 3 times less skew. Imagine how nasty the skew would be otherwise. Ouch.

So, they need at least 3 times faster fps, at least 3 times less skew to equal what they have now, and a lot less skew to go better than that, and something more clever than a simple fast buffer for the extended use of the high fps. And that's just the start of it.

Graeme
 
Graeme Nattress said:
Just to add - it's not just the fps, but the sustained fps. In raw or jpeg these cameras have 10-13fps for burst. They're still way short of the 24fps minimum hence the line skipping we see in the zone plates. Not only does this fake 3 times the fps, but 3 times less skew. Imagine how nasty the skew would be otherwise. Ouch.

So, they need at least 3 times faster fps, at least 3 times less skew to equal what they have now, and a lot less skew to go better than that, and something more clever than a simple fast buffer for the extended use of the high fps. And that's just the start of it.

Graeme
Click to expand...


Thanks for taking the time Gramme.

So are the sensors in these cameras being held back by something other than cooling (in terms of using the whole sensor at 24fps)? Such that if you built a regular style video body and used adequate cooling, you may be able to run them at that speed? I ask this mainly due to all the talk about Canon/Nikon/Panny slapping these sensors into a video body. If heat isn't a major factor, what else might be or do you simply have to design a sensor for video upfront if you want to be able to hit those FPS numbers?

If you don't mind answering something I asked about earlier. A lot of the talk about the poor performance of these DSLRs points to the OLPF being designed for high MP stills and not low MP video. You guys are clearly getting around this by just shooting your stills and video at the same res (assuming thing is an issue). Could you theoretically improve, at least to some degree, the video image quality of these cams by using an OLPF designed for 1080p sensors?

(and my apologies to asking what are probably incredibly stupid questions to somebody with your knowledge :redface:)
 
Contrary to popular belief, I'm not a sensor guru. I do know the odd bit about them though. From what I understand, getting very fast readout from a sensor is a non-trivial task. It's not just a heat issue, but that is an issue.

I don't think you could design an OLPF for the video sensors due to the line skipping giving an effective 1/3 fill factor vertically. Low fill factor approximates closer to point sampling, which needs even more OLPF than a fuller fill factor. In the end, you'd have to reduce the resolution so much to fully stop aliasing there'd not be much resolution left.

Graeme
 
Barry Green just had a thread over at dvxuser that shows the week link in the 7D, it is not the sensor, it is the hardware in the cam used for processing the image.

Lots of room to improve this cam without touching the sensor.
 
Chip G said:
Barry Green just had a thread over at dvxuser that shows the week link in the 7D, it is not the sensor, it is the hardware in the cam used for processing the image.

Lots of room to improve this cam without touching the sensor.
Click to expand...

I don't believe it. Sorry.

Jim
 
Chip G said:
Barry Green just had a thread over at dvxuser that shows the week link in the 7D, it is not the sensor, it is the hardware in the cam used for processing the image.

Lots of room to improve this cam without touching the sensor.
Click to expand...

The 7D has a great sensor for stills for sure. As for use on video, that's both a sensor issue and a hardware issue to deal with the full data off the sensor. Both need to be in place to see better video.

To me, the weak link come from the line skipping. Doing post-binning of the raw sensor data does not take vast amounts of hardware processing, so I'm sure they'd have done that if they could read the full sensor in real time. That means it still looks to me it's not a sensor designed for video. The amount of skew in the sensor (taking into account the x3 improvement through line skipping) tells the same story.

Graeme
 
Zac C said:
Jim, Barry did test it, and yes its the processing that Skips Lines...

(So you and Chip both have half the right answer)

It skips lines because it can't process fast enough
Click to expand...

Sorry. Wrong. The native Canon sensor will NOT run at 24fps.

Spin it any way you want, the sensor is not fast enough.

Jim
 
Zac C said:
Jim, Barry did test it, and yes its the processing that Skips Lines...

(So you and Chip both have half the right answer)

It skips lines because it can't process fast enough
Click to expand...

Looking from it from the point of view of someone deeply involved in designing cameras.....

There's no point reading lines off the sensor that you're going to ignore in the processing. And given skew is an issue, there's no point having 3x the skew for no visual benefit by reading all the lines just to ignore them.

Conclusion - the only place it makes any sense at all to skip lines is on the sensor read as this at least gives you 3x less skew than reading the full sensor.

Graeme
 
I wasn't agreeing with anyone either... the sensor on the canon can't do it, so they have to cheat it...

so maybe I was agreeing with you, but caring for both teams
 
Never mind, I know you have better things to do right now.

Sorry for bringing it up but yea, no audience member could tell the difference, however they complained about sunlight shots even with IR filters. Go figure.
 
I am curious.

I am curious.

I am interested in the OP's question from a different direction.So how about we say good night to high frame rates for a second ( as important as they are) and just say - what if we tapped the raw sensor data from say a GH1 which is running at 24 fps already and captured that?

Where would the difficulty lie?I understand high frame rates but what about bog standard 24p?

I can imagine tapping it is problem one - having a high bandwidth port is problem two and what format to write it too is problem three.

Tapping it is an issue, depending on how the sensor reads whether it is the whole sensor or line skipping,binning etc - a port is an issue again but I can see Cineform being an excellent codec to store it too.

But wouldn't this be an excellent way to get great HD - like the Andromeda system ?
 
I dont know if anyone can remember what a jpg still from a nikon D1, or 6mp D100 RAW file looks like

Very very nice

It would be interesting to explore the options available using lower resolution chips but pulling their RAW data

Judging by this thread it seems that pulling RAW from a DSLR is no walk in the park

Although we still have not seen conclusive reasoning why it isnt a generation or two away if it is just read write speed

A 24p RAW 6mp S35 chip should be possible if someone wanted to develop it ?

While I love 2k or 4k concept a really solid 1080 RAW would be good for me

I get a real wooly feeling when I watch any of my footage, 7d, EX1, I think the 5d was better but I flogged it

Hey Jim get on it with a decent 1080RAW solution at a budget ?

S
 
morgan_moore said:
I dont know if anyone can remember what a jpg still from a nikon D1, or 6mp D100 RAW file looks like

Very very nice
Click to expand...

10 years ago it wasn't bad, but they can't hold a candle to today's cameras.

morgan_moore said:
While I love 2k or 4k concept a really solid 1080 RAW would be good for me
Click to expand...

Personally, I'll take as much resolution as I can get.
 
Daniel Browning said:
10 years ago it wasn't bad, but they can't hold a candle to today's cameras.



Personally, I'll take as much resolution as I can get.
Click to expand...

I think those stills still hold a candle to a EX1 soft or 7d (alised and soft) frame grabs

Of course resolution is good but realistically Im in the market to expand my still photo offering with attractive (read S35) corporate shorts and decent 1080 is good enough for that and wont pay for a R1 which of course Id like but cant justify

S
 
You must log in or register to reply here.
Back
Top