Wolverine-Hawkeye Telecine

[Stan Jelavic]

Check this Kamel:
https://hackaday.com/2017/12/15/ther...d-3d-printing/

[Kamel Ikhlef]

Thanks Stan for the link.
This mainly concerns thermistors for hot end. But this is very very interesting.
I did a solder on a power pin (looked cracked).
It works again but I have the impression that the temperature is rising more slowly than before.
We'll see...

[Stan Jelavic]

That is good news Kamel. Hope it works for you.

[Kamel Ikhlef]

Hi all,
I hope everyone is doing well.
We are in confinement for several weeks in France.
Let us hope that this ends and that we can resume a normal life.

Here is the progress of my design. I chose to do in 2 parts to simplify because it is not easy to adjust the elements.
This weekend I continue on the upper cover by providing different camera configurations.

It advances ...

Take care of yourself in these special times.

[Kamel Ikhlef]

Finally 3 Parts :

[Stan Jelavic]

What is the green part for Kamel? Cannot see the whole thing.

[Kamel Ikhlef]

Hi Stan,
Glad you are well.
it is a removable magnetic door which will protect the film from dust and stray light during the scan.
But this still requires adjustments. It's not finish, it's quite long.
I take advantage of confinement to move forward.

[David Brown]

Always good to hear that Kamel's printer is working!

I used a bar code cut to fit onto 8mm film. Not the ultimate lens test, but works for now!



A case for upgrading the sensor to 3k or 4k.

Using the full resolution of the sensor for image data was always a no brainer for me. Well, now I know why it's not done. Framing the image is not possible without a physical adjustment for the sensor in X & Y. Drilling oversized mounting holes of the circuit board allows permanent framing for S8 but not for R8 in Y. Each has a different image/perforation relationship.



Capture all the R8, then adjust for S8. That's where I am now. A tedious process to change from one format another.
That is even more frustrating if there is no perceptible difference in image quality! This is more about preventing resolution loss when the final image is upscaled.



I'm capturing at 2792 x 1944, but that might be cropped to 1600x1200, then scaled to 1440x1080. That could then be upscaled to 4k (3840x2160) by the playback device. It's important that the original capture resolution is >= the source. That's lp/mm not the physical sizes.

Digital color sensors are really 30% less than the stated resolution and film is hard to calculate.
3k or 4k sensors would offer digital framing and maybe survive the crop and upscale. Even 2x binning becomes possible!

That being said, Stan's final test captures looked pretty good on my 4k monitor!

[Bruce Davis]

Kamel, appreciate the work you putting into this, it can not be easy.

Stan, wondering how the toothed pulley turned out from Shapeways? Fitted the S8 capstan Back pulley and gently seated each set screw on the motor shaft, once they were both in contact with the shaft gave each set screw a little more tightening. Fitted the sprocket and front pulley, placed a dial indicator on the front face and the wobble was 3 thou (0.003 inches), that's not bad.

David, are you still using the DFM 72BUC02-ML camera (2,592×1,944), or have you upgraded. I can see how using the full sensor area would be problematic re-alignment. If I understand correctly S8 film only has a horizontal resolution of around 900 lines. Capturing at double that, say 1500 to 1800 lines should give very good results (can not capture what is not there), but higher resolution is better for post processing. With the standard lens from Stan, once framed I end up with about 960 lines or horizontal resolution.

David Quote "I'm capturing at 2792 x 1944" is that the framed resolution or full overscan, or are you using a different lens from mine (probably).

Chris, how is your project coming along, looks like we will all be in lock down for a while with Coronavirus worldwide.

Regards - Bruce

[Brian Fretwell]

Rather difficult quoting a resolution of film in lines when talking about this sort of work as films used to quote it in line pairs/mm and digital in display lines (originally scan lines on a CRT) which is half the film version. I always get confused as to which system people are talking in.

[Stan Jelavic]

Should look pretty nice once done Kamel.
Good work David. Which lens did you end up with? There is a bit of blurring at the edges from what I can tell but not sure. It would be nice to compare the final result between the 12mm and 16mm on your monitor.
Hi Bruce, I did receive the toothed wheels from Shapeways and look good. Will do more testing today.

[Kamel Ikhlef]

Hello Everybody,
Here the Front cover Bottom STEP File, if you want testing it.
I continue for the upper cover and the magnetic door.

My Printing settings :

ABS
With support.
infill 50%
layer 0.15mm
Temp Bed 90°
Temp Hotend 230 °
Speed 50mm/s

[Chris LaRocque]

Hello All

Shelter in place is upon us all in the Eastern US. I've been allowed to work from home for the duration and am fortunate to have fast fiber (FIOS). The last week has been all about securing inlaws / parents and bringing all of my equipment home. The dining room table is covered with computer, circuit boards, meters, scope, and power supply. Keeping the cats off the table is a challenge.

Stan Jelavic
About having the sprocket holes in FOV. The camera my father used didn't provide a distinct frame boundary. Is it common for the image to extend past the sprocket holes? I suppose this was a "cheap" camera? In the example frame I posted earlier, the vertical frame boundary opposite the sprocket holes is not contiguous and the top edge of the lower frame line is angled. My thought is to use the sprocket holes as a reference to validate the top and bottom frame line. This will also allow vertical alignment correction using the inner edges of the holes.

Bruce Davis
As for autocrop progress. I've put the effort to repair frames that move out of FOV on hold. I'll implement it in the future but for now I'm focusing on images that are fully in the FOV. The current code reliably picks up the sprocket holes and uses their inner edges as the left frame boundary so I don't have to estimate where to place it. Of course if there is a good frame boundary then it will be used preferentially. The upper / lower frame lines and the opposite non-sprocket edge are distinct enough that a reasonable boundary line can always be selected. The code detects all lines and filters out any having a slope greater than 0.2%, then uses the detect contours function to select rectangles having an aspect ratio between 1.2 and 1.5 and an area that is greater than or equal to 70% of the total image size. I need to add some code to detect and filter out lines from the central part of the frame and test the upper / lower detected frame lines against the sprocket holes. I want to use the pixel distance between the holes as a scale to determine the midpoint of the holes defining the frame lines on D8 stock and frame center for S8.

Moving forward I'm going to try using the first detected frame boundary as an ROI for one of the tracking functions. I hope it will reduce the probability of introducing frame jitter due to uncertainty in the edge detection algorythm.

I've only tested on single frame images so far but have it set up to accept a directory full of frame files or a list of files through stdin. I'll eventually set it up for operating on a live video stream.

Hope everyone else is doing well and staying safe.

Regards

Chris

[David Brown]

Chris
I've seen telecines that are using sprocket holes for registration and have no claw.

http://8mmforum.film-tech.com/vbb/fo...=3875#post3875

The link above will give you dimensions, use the long pitch, it was for reversal film. I have R8 films that have an image across the whole width, except for the sprocket holes.

With S8, the frame lines can be thick and black or crazy thin. It's the camera used that determines thickness.

Bruce
I am using the DFM72BUC02-ML sensor spec'd by Stan with a 16mm lens.

https://www.aico-lens.com/?s=ACH1656MAC

1piece ACH1656MAC f5.6 with IR cut filter will be 3 weeks after receiving payment. The total cost for 1piece ACH1656MAC is $115($85+$30 by Fedex/UPS to United States).


These are 20mm standoffs. I have other sizes that are too large to shorten just a little. The Wolverine cutout limits the total lens drop with out modifying the face plate.

I was capturing with the full size of the sensor 2792x1944. It is not useful to do that. I had hoped for a little less FOV from the lens. In the end the final crop will be more like 1600x1200. There can be fine tuning with the partial scan when less than full sensor size.



I have read an account comparing Kodachrome to Ektar negative film in the 90's. Kodak was advertising Ektar to be 200lp/mm, and better than Kodachrome. Someone who shot both stocks on the same equipment and has access to high end microscopes, claimed that they were equal.
One reason film resolution is hard to calculate is because there is no information on the lens used. If the lens can't resolve 200lp/mm the film can't record it. I don't remember anyone comparing Super8 lens quality except thru reputation of the manufacturer.



Brian I think the LP/mm is not used much anymore. The pixel dimensions of a sensor or crop don't tell you the pixel pitch which more closely matches the LP/mm.
Now lenses have MTF charts or a megapixel ratings to match with the sensor.

It says right on this card "for reference only". I though it might be useful.
This is literally 1 line pair per mm. Try to imagine 100 or 200 LP/mm !!!

Stan I have more time to spend on this lens thing. It's been a few minutes here and there. I'll try to remember the standoff size for 12mm.

I would much rather have used a long 2mm bolt and hollow spacers. These are threaded at both ends. Very cumbersome.

[Kamel Ikhlef]

David,
I'm doing the top cover design.
It would be interesting to have your dimensions with a 16mm lens (20 / 22mm spacers?)
An image of your installation would be good.
I will take into account the size for your configuration.