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 :)

RED and SI?

I wouldn't worry about that. Because there is noise in the image and you already have 10-bits, it is highly unlikely that you'll end up with banding issues from the acquisition source. (e.g. think of people shooting on 8-bit formats like HDCAM, DVCPRO HD.)

2- Thinking aloud here... the SI-2K has less optical low pass filtering than the Red does. You might end up with aliasing artifacts (these look worse with Bayer) that don't show up in the Red footage. Perhaps this could be an issue???

*Disclaimer: I haven't tested this to see how objectionable the difference would be. Some diffusion filters on the lens would probably solve it.
 
their OLPF is tuned for 2k shooting. Red's is tuned for 4k shooting. Greame spoke to that point in an older post. Technically Red is inferior for 2k shooting because the OLPF is set up for 4k scaling and display. Ie when blown up to the same physical size the olpf on the red is meant for a magnification factor based on 4k, and the Si is made for one based on 2k. just like 16mm vs 25
 
Technically Red is inferior for 2k shooting because the OLPF is set up for 4k scaling and display.
Click to expand...
No it's not. :) The 4K oversampling approach gives you the most flexibility in making good 2K images.

When you sample a signal, you have to essentially pick your poison:
- Imperfect frequency response / blurriness (not a problem if your sampling rate is sufficiently high)
- Aliasing / moire
- Ringing artifacts

Ideally, you'd pick the artifact you find the least objectionable (probably ringing). By shooting at 2k for 2k output, you are stuck with whatever balance of frequency response + aliasing + ringing the camera does. The SI-2K shouldn't have any ringing artifacts (because it's not possible to achieve that kind of filtering optically).

2- With Red, when you rescale from 4k to 2k, you are resampling the signal. And by choosing different rescaling algorithms, you can adjust the balance of frequency response / aliasing / ringing that you get. So that aspect would probably work.

2b- If you had shot 2k windowed on Red, then I'd think that you'd find that Red aims for less aliasing at the expense of frequency response / sharpness.

2c- Back to 4K scaled down to 2K.
What won't match by default is how the aliasing artifacts look. When you have aliasing in a Bayer pattern sensor, you get color artifacts where there shouldn't be any color. e.g. if you shoot a black and white zone plate pattern, the resulting image will have colors where there shouldn't be any.

Whereas if you shoot 4k on the Red and rescale to 2K and try to introduce aliasing artifacts, that aliasing from the rescaling will not have the same look. The end result is that the SI images can have color artifacts on black and white objects like newspapers, while the Red 4K-->2K images won't.

*Non-ideal debayering algorithms can also cause color artifacts on black and white objects... this is a different issue. Apparently SI has updated debayer algorithms for their camera and therefore the examples on their site aren't as good as they could be.

3- You can get an idea of what I'm talking about by looking at some Red images and the SI examples on the website (but don't look at anything with chroma subsampling... e.g. not the windows media player movies).

In particular, look at objects which you know would be black and white in real life. In the SI images, you can see saturated colors where there shouldn't be any. Does that bother you?

4- This may not be an issue in practice! And other issues might be more important... e.g. noise, the color reproduction of the camera, etc.
 
OLPF's are not set for any specific resolution ever. Rather, they are designed for a specific photosite spacing. The resolution that results is a matter of the size of the array, but the properties of the OLPF are specific to the inter receptor spacing exclusively.
 
I was just reiterating a comment made previously by Graeme. And yes thank you David I was referring to 2k windowed, not 4k scaled to 2k.

Kevin I have no idea about the pixel density or size of the Si. I cant imagine it differs to much from the Red considering the sensor sizes. Graeme said they had picked a rather strong filtration for Red's chip, so take that for whatever it means. I have heard him speak to it in far less technical terms, and regarding the amount of magnification that is intended for the end result. Similar to the Circle of Confusion concept for DOF.


Also GlennChan: calm the hell down. I was speaking only about 2k windowed, you are right a 4k image scaled is far superior.
 
I don't know what comment from me you picked up on Mike, but KHMuse has it right on the OLPF stuff. The OLPF is set for a particular spacing of pixels. You make those pixels bigger, you increase the spacing on the OLPF also.

Graeme
 
It was a while back, I can't even remember what the thread was really about. We might have been discussing aliasing artifacts or Nyquist this or that.
 
No worries.

I've not seen any SI 2K shots of a zone plate, so I've no idea how it behaves with regards to aliasing or how it's OLPF behaves.

Graeme
 
Also GlennChan: calm the hell down. I was speaking only about 2k windowed, you are right a 4k image scaled is far superior.
Click to expand...

Hey, I didn't mean to disparage you or put you down or try to get into a pissing match with you or anything like that. Sometimes it's hard to convey tone on the Internet you know what I mean?

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