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 :)

Ikonoscope A-cam dII

DanGaud said:
hehe i know, but in Redcode 28 the blue channel looks more like a cheap fractal texture than actual grain.
Click to expand...

I agree, I don't mind grain and noise that's plain noise, but when it's compression noise it gets the artifacts of the compression and becomes difficult to handle with. With 2K it all becomes a lot more noticeable. This will all be different when the new sensors come which has less compression.
 
Hey Matt, thanks for interview... headed there to listen to it now.

BTW, I sent you a PM just now.

Take care man, hope all is well in Fayetteville.
 
Test footage from A-cam dII for download

Test footage from A-cam dII for download

A-cam dII test sample footage is up on their site for download as h.264 files.

There are several issues with the test footage:

1) Where is the tripod?

2) What's with the lighting?

3) Skin tone is off balance? Need better lighting and chroma matrix?

4) Looks like the shutter angles are set wrong?

5) de-Bayer needs work, no DNG sample frames for me to play with :(

6) Are they using a OLPF? Seems too sharp or there are alaising issues, even when viewed 1:1?

Other than that it looks like it works, much sharper than 16mm, I would say 16mm is dead if they get the dII a little more polished... maybe?

Question:

Does anyone know a good free Windows program that can make h.264 video files from a set of 24bpp BMP and or 48bpp TIF frame files?

And also is there a free program that can break up a h.264 video file into a set of 24bpp BMP and or 48bpp TIF frame files?
 
I'd heard they were using an OLPF, but those shots don't look like it. There's what looks like aliasing / chroma greeblies going on in the watch shot.

Graeme
 
greeblies?

greeblies?

Graeme Nattress said:
I'd heard they were using an OLPF, but those shots don't look like it. There's what looks like aliasing / chroma greeblies going on in the watch shot.

Graeme
Click to expand...

Is "greeblies" an official term? I was wondering what those are called, I looked at 2x and 1x and they seem to be in the file and not a problem with the viewer program on the computer (my Brother's computer only goes to 1280x1024 now so I was looking at part of the image).

I think you can filter those out with a software filter, well to some extent if the lens is not sharp enough to just spot a single pixel, but it uses more processing time, 5 minutes per frame would seem to be a limit for total processing, CC on 1920x1080 frames is less than 30 seconds, general copy time might be 2 minutes, so that leaves maybe 2:30 for de-Bayer?

Anyway, it seems they have it working and expect to be shipping around NAB 2009?

With your new sensors the pixels will get smaller, so at some point will making a OLPF with small enough divergence become an issue, or since the lenses will not be able to focus on a single pixel will the OLPF not be needed?
 
Greeblies is an official Graeme term. Without looking at the raw data, it's hard to know what's going. I hope they can present more and improved movies.

Yes, if you have small enough pixels that the mtf of the lens stops enough of the highest detail coming through that you have little or no aliasing, then indeed, an OLPF is not needed.

Graeme
 
Charge Couple Device

Charge Couple Device

Yep, the Ikonoskop dII is CCD based. We really could have used global shutter on a particularly gnarly fast dolly move this past week...
 
Just because it is CCD does not preclude a Bayer pattern mask. Global shutter or rolling shutter, bayer or striped mask, binning, single chip or 3-chip with prism block, and so on. They can all be applied to CCD or CMOS.
 
Gunleik Groven said:
ok???

Looks good, but I cannot really see that last sentence from the images...

Body/CF module: 18
EVF: 3
10 CF cards: 2 (40minutes of recording...)
Lens on eBay: 3

That's 25k
and then?
What makes up for the next 25?
Click to expand...

RED ONE body
Base Production Pack
2 RED Power Packs w/ 2 batteries
RED LCD 7"
RED CF Module
RED CF card (5) Get 2 - 16gb CF cards instead and check out the ET shock drive mount that makes the need for using CF cards less necessary.
RED EVF with the ET (Element Technica) EVF mount
RED Case
Nikon Mount
RED Drive x 2
Drive Cable x 2
Mini-xlr to XLR cable One comes with the camera but you will need at least 3 and 4 would give you an extra for backup
Cheese Plate Get the ET top cheeseplate or see below for the ET handle
Red Quickplate
Red Handle (left)
Red Handle (right)
Side Handle
Red Handle Mount
Red Arm
Red Wingnut
Bolt Set
15 mm Adapter x 2
Focus Hook Set
Top Handle
Top Handle Extension
Universal Mount 3 or even 4
Steel 6-inch rod (2)
Carbon X 18-inch rod
Red Brick - 2 more to add to the 4 that come with the Power packs.
AC Adapter
Tripod. O'connor 1030
Lens: RED 18-50 T3, Zeiss ZF 18, 50, 85

Total: around 35-40k

Ikonoskop: Max USD 13k (with IMS Mount and shipping)
 
Artifacts

Artifacts

Kenn Michael said:
They're using a CCD not CMOS sensor so no debayer?
Click to expand...

It seems they are using a 1920x1080 Kodak sensor KAI-02150, if not someone can point that out...

http://www.kodak.com/global/en/business/ISS/Products/Interline/index.jhtml?pq-path=11937/11938/12032

http://www.kodak.com/global/en/busi...erline/KAI-02150/overview.jhtml?pq-path=13614

http://www.kodak.com/global/en/busi...ine/KAI-02150/specs.jhtml?pq-path=13614/13615

2/3" C mount Megapixel Machine Vision lenses should work with it?

The image circle is 12.1mm.

Maxes out at 64fps?

The ordering options page shows,

http://www.kodak.com/global/en/busi.../KAI-02150/ordering.jhtml?pq-path=13614/13616

Bayer RGB as the Color option. Kodak (tm) was working on a sensor with one green and one clear filter (rather than two green as in the Bayer), but that does not seem to be on the list.

Anyway, from the typical artifacts in the Watch image you might think it was a poor de-Bayer without a heavy OLPF. They may be using the free DNG de-Bayer program, camera makers sometimes have their own de-Bayer programs like Canon to get better results from their own camera. Several values must be tuned to match the sensor, light color values, lens and OLPF in order to get artifact "free" results. Its could be best to use a program tuned to the camera, thats why the R3D files only go through RED developed code for now (there is no RAW data from the RED ONE, and you need to use their SDK to de-Bayer the REDCODE compressed sensor data.) The DNG de-Bayer is sort of a generic SDK, not for a particular sensor, lens, lighting, and OLPF any of which can unblanace the de-Bayer, it has values to adjust it but the code used is not changed (or optimized for the camera?).
 
Dec!

Dec!

Matt Garrett said:
They said orders sent in NOW will be filled by april. Not sure on when they plan on shipping the first preorders.
When first announced some preorders were to be shipped by dec. and the next batch early jan.
Click to expand...

Well, if its out in Dec. we should see some more footage soon!

I wonder if they will set up something like REDRELAY for people to work with the user's DNG files to see how the true RAW Bayer data workflow fits into their editing system and TB storage.

It looks like they have more than a few weeks of work to do yet on both the software and the backfocus adjustment (do you see a backfocus adjustment on the camera for field adjustments, the "plate effect" throws off the backfocus for various lenses (especially fast ones), so you need to adjust the backfocus if you want the marks to get closer, otherwise they will have to set it so some lenses go past INF in order to be sure you can get sharp INF focus?

If the OLPF comes out you need even more range on the backfocus adjustments.
 
Sounds like the shots they posted were without the intended OLPF in place, hence the aliasing and chroma nasties we could see.

Graeme
 
You can usually reckon on about 70-80% approx of the linear pixel dimension as a measured resolution.

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