Kamel: I second that.
Announcement
Collapse
No announcement yet.
Wolverine-Hawkeye Telecine
Collapse
X
-
Everyone Here's a couple of tips.
Control capture file size with ROI (region of interest) cropping. Just stumbled on to this. Cut my files to 1/3 the size.
It's part of IC capture.
I am cropped pretty close intentionally, but not as much as I would for a final video. This was another capture test. I'm using an old Castle Films print. I'm surprised how soft the image is! I can see the film grain, so I'm in focus.
This could be a warning sign of jitter. Watch the film between the gate and the first pulley. If it's curved, not straight, there is no tension back to the gate.
https://drive.google.com/open?id=1kd...sBmyvxdVCoQ1Nm
The capstan (smooth pulley) is stuck and not rotating causing some friction.
And here's the capture.
https://drive.google.com/open?id=1jc...qLJPzH7iv-JIHpLast edited by David Brown; March 10, 2020, 10:54 AM.
Comment
-
Kamel, sorry for the confusion regarding 3.8mm and 3.5mm, even further confusion Stan's measurement is also different to both ours.
David, I have been using ROI but due to jitter can not frame as accurately as you have. Your video is excellently steady frame to frame.
Link for Capstan printed parts, these will suit 3.8mm or 3.5mm from the metal support face, if for arguments sake yours varies from this, say 3.2mm, just edit the step file and remove 0.3mm of material from the back face of the "Back_Pulley_Capstan".
https://1drv.ms/u/s!AtIUdJCJJJGigRH5...kvsH_?e=iZERrB
Regards - Bruce
Comment
-
Hi
Thought I'd share progress. This is the original frame, the detected lines, and the final crop to a 1.33.1 aspect ratio referenced to the right edge of the frame. I'll have the code able to process a directory later this week.
I'm considering masking the corners to simulate an idealized frame and thinking about how to deal with frame creep; opinions?
David, is that contact sheet you put up yesterday only an illustration or are those captures? If caps can I get the series to test partial frame buffer techniques?
Regards
C.
Comment
-
Looks like David and Chris are doing some exciting stuff. A few comments:
- Good input on ROI.
- Can you provide some more data on image sharpness with the 16mm David. Hard to say from your clip. Will have to do the hi rez frame grab and check the corners.
- Good work Chris. Do you plan to use manual exposure with this cropping method?
Here is my capstan test. The capstan pulley was slipping and had to tighten it up and ended up crooked, Still worked:
https://photos.app.goo.gl/V9gNEzYNLPk8rUtf7
Comment
-
Stan Wish I had the Wittner test film ($80), I bought a couple of commercial prints for S8 and R8 thinking they would be sharp. I will try to find a good subject for sharpness testing. I have been hung up on jitter.
Bruce I'm just as amazed about the tight crop and steady capture!
Chris I will upload those tomorrow.
Kamel Could you lengthen the "H22_Support_M12_Lens_v25" to 26mm?
I'm going Ice fishing Thursday and have fallen behind in preparation! I will be gone thru Sunday.
Comment
-
David, Here is the 26mm M12 Lens Holder STL file:
https://drive.google.com/file/d/1DsJ...ew?usp=sharing
For the wittner clip. I found a small piece of about 10cm. This may be sufficient for testing.
I can send it to you for free if you want by ordinary mail
(give me your address by private message).
Comment
-
I like the turn that events take !
David experiments,
3d model of Bruce,
Chris automatic cropping,
V12 update of Stan with improvement of the jitter ...
Very interesting! It makes me think that I was way behind the front cover.Last edited by Kamel Ikhlef; March 11, 2020, 06:36 AM.
Comment
-
Kamel
The images I'm working with were acquired on my unmodified Wolverine Pro with default settings: 1920 x 1080. I had tried a close FOV and decided to pull back to see what it looked like. I suspect that the 'compression' artifacts of the original Wolverine are really interpolation artifacts from digital zoom that might be resolved with a better lens; but no matter. Final crop is ~1276x960. The dfm72buc02ml will do 6 fps at 2,592×1,944 so I will crop then resize down to 1436x1080 and pillarbox the result for display on 16/9 screens.
Stan
I had thought I'd let it auto expose and then correct the raw image before debayer if necessary.
Stan, how hard would it be to pulse the trigger multiple (odd) times per frame? I was thinking about median stacking like the Astro guys.
Comment
-
Hi Chris, the dfm72buc02ml camera captures the image at 12bit and then reduces it to 8 bit. We do use the WDR feature to improve the dynamic range but still the exposure has to be right to prevent the image from hitting the low or high side clipping. Post processing can correct to a certain extent but once the info is gone it is hard to compensate for that. That is why I mentioned the auto exposure that will be on a short side due to perforations.
Why not use custom centering so that the holes are out of the view and use black film edges for alignment.
Or you can use your own white balance and use camera api. Nick wrote the code that utilizes that api and if you contact him I am sure he is willing to share the source. I compiled the code with Visual Studio and it ran ok but did not modify anything.
Yes, multiple triggers are quite feasible with the V12 board. Some MSP fw change would be necessary.
Comment
Comment