Wolverine-Hawkeye Telecine

[Stan Jelavic]

No problem Bruce. Will send you the MSP. The V10 supports capstan at 1 FPS but not at 1.6FPS. Wolverine runs at 2FPS.

[Kamel Ikhlef]

Hello everybody,
Stan made the good point that it was not easy to navigate in all my 3d files to print on thingiverse.
I admit that I myself had trouble finding my way.
But on thingiverse it is not easy to classify well. So I reorganized all the files in a Zip archive.
Link googledrive : https://drive.google.com/file/d/1gin...ew?usp=sharing
Link Thingiverse :https://www.thingiverse.com/thing:3891738
Have Fun

[Stan Jelavic]

Thank you Kamel. This makes the things easier everybody.

[David Brown]

Thanks for the suggestions Stan. I'm not certain what did the trick, but my money's on the driver. Possibly using previuosly saved configurations didn't help after I had the right driver installed.
I was missing more than the 2 highest capture resolutions. Options in the software were missing or grayed out, like gamma and more. FYI to all who have this problem here's what's installed.



I thought of staying with 1/2" lens or even 1/1.8" lens to keep the same mounting in place. Those sizes are not available in 10mp, just up to 5mp. Oh well, the 2/3" sensor diagonal is only 3mm longer.

Since the lowest resolution I can use without clipping the image with this 16mm x 1/2" lens is 2048x1536. I was able to center X but there was not enough room to center Y. So I moved the circuit board to center the frame. Held on with clamps right now. Small offset, could drill the mounting holes eccentric. The screen capture is set to 2592x1944. I could leave it there, do pixel binning, or another method to hit 1440x1080.


Lens is in Kamel's 22 mm mount and I used 20mm standoffs. That will change with the actual lens.


Stan, looks like the shutter will stay open for 30 seconds, does V10 allow for the shutter to stay open for most of the cycle of the frame rate if needed?

Now it's time to connect the camera to Hawkeye. I still don't know if I have fried the opto coupler!

[Kamel Ikhlef]

David, If you need a specific camera standoff (a specific higher or lower size).
I can design a 3D part, It will be usefull for testing). Tell me the size you would need.


I designed a 3D PCb Test to check the holes and the position of the rear USB to be sure to center the camera and be able to close the back cover of the telecine.
It seem Mike was forced to trim his hood to pass the rear usb connector.

you can also adjust the offsets through the settings of the iccapture software to crop.
But I don't know if this procedure influences the final quality of the image capture. I Think It is better to be as centered as possible.

[Stan Jelavic]

How is the image quality David? The board does not put any constraints on the frame rate since it just passes the frame sync from Wolverine into the cam. But the way the wolverine frame sync is done could be a problem because the sync is at the beginning of the cycle instead of the end.

[Stan Jelavic]

Here is the frame sync mod that puts the sync at the end of the claw cycle. That allows plenty of time for camera to take the image without smear.
It also may be possible to bump up the FPS because the claw delay implemented currently is no longer required. Minimal change to the BOM but would have to go with V11.
Obviously, will test first. The change is pretty simple.

[Bruce Davis]

Hi Stan, firstly thanks very much for providing a new MSP430 to improve my capture speed.
If it is possible to improve capture speed to 2FPS I would be interested. I do not understand the technical part (your schematic above) and how you would achieve this. You previously mentioned that a trigger sensor would have to be re-located and that would be a considerably difficult modification. Thanks for your ongoing dedication to this project.

Hi Kamel Ikhlef, thanks for providing & re-organising your printable parts into separate folders, very nicely done.

Hi David Brown, would also like to thank you for your work with optional lenses, in order to improve frame capture resolution.

Regards - Bruce

[Kamel Ikhlef]

Stan,

With this very, very interesting improvement (4K capture forecast or it will take more time for each capture),
should the MSP code be modified or may it have remained as it is currently (last genfreq8 version)?

Another thing that would be really really great, but I do not know if this is possible :
It is to offset the winding of the film on the takeup reel (offset the synchro of the takeup motor).
for example run the takup motor all the every 10 or 20 images recorded to avoid the big problem of jig encountered by everyone.
(the solution of folded paper is not really "professional".
But I am not competent enough in this area to know whether this can be done or not.

[Stan Jelavic]

I have not shipped you the part yet Bruce but will do this week. Been a bit busier at work this week. Here is a simpler explanation of the frame sync mod. I do not plan to do any mods to the sync mechanism just intercept the sync in MSP430 and delay it sufficiently to get delivered to the camera at the point where the claw pulled the film all the way to the right and starts going down. So now the camera has more time to take the picture while the claw is retracting back to the left. Currently I am halting the stepper by slowing down the clock right after the sync so that the image does not get smeared because there is not much time between the sync and the film movement (all done at the beginning of the claw cycle). But hating the clock at any point impacts the FPS...
Here a small pict explaining it. Hope it makes sense.

The new design will require a firmware mod to handle the sync intercept.

Kamel, I do not see any jitter from the takeup with the latest design. It is the same as no takeup at all. From what I can see, the jitter is introduced by the film itself and the way the claw engages with the holes and also from the gate itself. Option 4 of FredVideo script gets rid of the jitter almost completely.

[Bruce Davis]

Stan thanks for the clarification, if I understand correctly the triggering will then always be 1 frame behind, in other words the very first frame is missed and the second frame is captured with the first frame trigger (due to the delay). I feel that the actual triggering would need to take place after the claw is free of the film perforations due to the pressure on the film while the claw is still engaged (this cause the film to be slightly out of being perfectly flat).

No rush with the MSP430 whenever it is convenient for you Stan. QUOTE "Kamel, I do not see any jitter from the takeup with the latest design", is this due to a firmware change or a hardware change. I use VideoFred script to get rid of what you are calling jitter (assume you mean every frame capture is in a slightly different position causing an effect like camera shaking, e.g. no Sony steadyshot applied). Another option would be a VirtualDUB2 pluggin named "Deshaker v3.1". https://www.guthspot.se/video/deshaker.htm

EDIT - just thinking how is it that the original Wolverine camera is able to capture at 2fps in the limited time made available by the poorly designed trigger location and the DFM 72BUC02-ML can not (with darker frames), is the Wolverine sensor much faster. I assume they are both rolling shutter but perhaps the Wolverine sensor is capturing at a lower resolution (640x480 or 720x540 ?).
Even though only a small capture area of each of these sensors is actually utilized am I right in assuming that all of the sensors' pixels are read for each frame captured.
I capture at about 1336x992 for S8 but that can vary slightly.
Hawkeye DFM 72BUC02-ML: 5.0 Mega Pixel (2,592 × 1,944) 1/2.5" CMOS sensor
Wolverine Image sensor: 3.53 Mega pixels (2304H x 1536V) 1/3" CMOS sensor
Regards - Bruce

[Stan Jelavic]

Your understanding of new frame sync is correct Bruce. But also note that the delay can be adjusted so that the camera can be triggered at the point when the claw disengaged the film and is moving down. It will take some experimenting with the delay t get it right.
The jitter free (or image shake) is accomplished by reducing the pulse to the rewind motor. I tested it without capstan. I will test it once I get all the printed parts from
printathing.com. You can test it by running the scan with the rewind used and another scan by letting the film loose and manually rewinding or letting it drop into a box and then compare.

Mike reported some very good results with Deshaker:
https://www.facebook.com/mike.spice....7587802698181/

[David Brown]

Wow, you get sick for a few days and there is a lot to catch up on!

Stan, this is the most important need I have right now, the camera trigger. My optocoupler might be fried. First did I wire this correctly? From the Hawkeye manual V2 it show a provided sync cable. I had to make it. The picture has enough detail to select the camera side of the plug. The board side might be off. Here is what I did.


If this is wired correctly then I will try to buy a new OC from Wolverine.

I know you made some suggestions for testing voltages on the optocoupler, but I think this is the ultimate test. The external trigger is not working with the IC capture software. I can turn the timer on and it captures all day long. The procedure I remember, is to turn on the external trigger before starting the timer. Doing that, nothing is captured. This might means the OC is fried. I still have the peculiar startup problem. Unplug sync, plug in power, and then reconnect sync to start up.

Your question of the 16mm capture's quality, the image is soft, but the lens was $10. I think if I could set it at f8, it would be sharper. No option there! The lens I'm looking at is a 10mp, macro, & f5.6 its gotta be better.

Kamel do you have jitter in your captures? I had only one case that I found was on the original film, not caused by the Wolverine. I agree that the paper method is not a professional, but it does allow for variables in film thickness/width that can easily be adjusted. I placed a small strip, two layers thick, of "gorilla" tape in m gate that is close to the paper thickness. I did this to see if it is a catch all. This is not duct tape. It has a better quality adhesive and won't make a mess. I don't think of it as permanent.

Bruce at least the Hawkeye is capturing the cropped pixels you specify in the IC. My sensor is covered with dust. I found that useful to determine how the image is captured. I could see the increasing dust pattern with increasing resolutions. The partial capture moves that resolution in X or Y to center the film image. That won't work if your using a full resolution capture! Then you must move the board to center! I still have not captured any dark scenes to exceed the claw timing yet (since it's all done with the shutter), but I will. I know there are some scenes that will look bad no matter how long the exposure!

[Stan Jelavic]

I cannot open the pics David. The optocoupler test should be pretty easy. Connect a 100 ohm resistor between pin2 of the wolverine sync connector (using adapter cable) to 3.3V. Check voltage on pin 2. Should be around 1V. That is the optocoupler bias.
Then connect a 2.7k or similar value between pin3 and 3.3V. Make sure pins 1 and 4 are grounded. rotate the claw slowly or in bursts. The voltage on pin 3 should swing between 0 and 3.3V.

[David Brown]

I don't know what happened to the pictures. I would like to confirm the connections.

from the manual


as I wired it