Hawkeye-DSLR

[Stan Jelavic]

Hi Isidro.
Very good suggestions. The FPS should not be taken down too low. Around 6 FPS that you suggested sounds pretty good but would have to experiment with it.
Yes the extension ring would increase the resolution. Two concerns.
Could push the lens limits lowering the optical resolution i.e. increased pixel count but fuzzier image.
The front focusing distance with the extension is very close to the stepper motor. Adding an extension could lower that distance causing the lens to hit the stepper armature.

[Nantawat Kittiwarakul]

I'd like to share my 2 cents, as you may find it somewhat useful.

In my previous setup made of a miniDV camcorder and a stop motion capture software. That software would cherry-pick a still frame as per triggered by the sensor signal and stich them together as a single continuous video file at the end. I've found that it will work best if the trigger interval is the exact (or closest to) 1/n of the camcorder frame rate. After tweaking I've found that it would work best at 1/5 of the camcorder's frame rate. Running it any faster the trigger process will become too unreliable so this is the upper speed limit.
Since my camcorder is PAL it means 1/5*25=5 fps capture speed. So for a 50ft reel it will run through within 12 minutes and not even a single dropped frame as the result.

IMHO this should be applicable to HD/2K/4K setup as well - setting it to capture 1 in precisely every Nth frame of the incoming video stream. It would have to be some trial & error, and lots of tweaking to find the optimal "N". My guess is that the camera's high fps + fast PC + SSD hard drive would help considerably.

[Stan Jelavic]

That is a good idea for the setup where the video stream is routed to the computer. In my setup with the Mark2 the video is stored onto the camera card. I get extra images initially which I trash during post processing. It is slower than what you are suggesting Nantawat and it takes up more work storage but it also comes out very clean. No dropped or messed up frames because the Wolverine runs at 2FPS exactly and the camera FPS is a multiple of that. The advantage is that the setup is very simple. Do not need the camera computer connection and associated sw and cables.

[Stan Jelavic]

Got the Wolverine trigger interface to work today.
https://photos.app.goo.gl/9dUDMuQ4sgJg4AEh8

The picture shows the consisting of:
- Wolverine controller and power supply
- New interface board
The board intercepts the LED 3.3V power source, runs the board with it and returns it back to the LED. The on board MSP430 controls the camera trigger and the takeup motor tension.
The scope in the image displays the input trigger from the Wolverine controller and the trace on the bottom shows the output to the camera. As shown, the camera trigger is in the middle of the claw cycle.
The MSP430 also controls the takeup tension with 3 settings available from an external switch. That is displayed in the next image.
https://photos.app.goo.gl/XQPvAq9v3iaM4LxG8
Will put the whole system together tomorrow and run a short clip.

[Stan Jelavic]

Here the the setup running. Wolverine controller connected to the trigger interface board. Note that there are no wiring changes to the wolverine controller. Uses existing Wolverine cables and a few extra cables added. Note the pulsed takeup. There are 3 tension settings controlled by 2 switches.
Found out that I need a Wolverine run control switch needed for initial focusing adjustments. It is needed because you do not want the machine to run while adjusting focus but you still need the LED to be on. I also added the rewind and reverse switches similar to Hawkeye.
Some issues running at 2 FPS with the D5 Mark2 not being able to keep up at F16. Had to increase ISO to 2000.
https://photos.app.goo.gl/W1uduru41bDwPdk9A
The cropped video resolution is 3364x2500.

[David Brown]

Look'n good! I'm thinking this might also work with the microscope camera I bought. I'm pretty sure it can shoot video at 4k. The lens might not get close enough because of the stepper motor. I'll know later.

F16 seems a bit extreme. I have used an old Nikon lens (1960's) that is only sharp in the center. F8 would correct that. How much of the Cannon's sensor are you using? I doubt the 8mm frame fills the camera's frame. All this was from wondering how stable your camera support is. Is it fastened to the Wolverine's supporting mount?

[Stan Jelavic]

Would be interesting to see your microscope cam working. My problem is with the camera mount. It is a cheap stand and the clamp does not fit the lens very good so that I adjust it by wiggling it a bit and making sure that the test image is overall in focus. Will retest with the F8 settings. It is a bit fiddly. A good stand and clamp would resolve this issue.
The image covers around 50% of the sensor roughly.

[Stan Jelavic]

DSLR capture update.
Run another clip with aperture set to F8. Now, the camera leveling becomes trickier but still manageable. Set ISO to 1000. If too low the images get smeared because the camera cannot keep up. The format set to RAW1. White balance set to 5000 but does not matter much because can be changed in the post since this is the RAW format.
Imported the clip in daVinci. The preview looked very dark which is to be expected because the input is 12bit (or maybe 14 bit). So the bright sky is pushing the mid tones way low. The image looks purple. Again as expected due to the prevalent blue from the Wolverine LED.


I changed the FPS first. In edit mode (tab on bottom), right click on timeline and open up clip attributes and set the FPS there.
Next, increased the gain to be able to see the image properly. Click on color wheel on the bottom. Set the gain so that the image is clearly visible.


Next, turn the blue down until at roughly the same level as other colors on the scope.






The image blue hue is now gone...


Turn the combined gain back down and get the normal image back by manipulating the curves (curves button)
Go through these settings back and forth checking different scenes until satisfactory.
Note that the sky blue color will not be good even after all these settings are done. For that you have to do selective color (eyedropper icon).
Move to the scene with the blue sky.

Create new node and link it in.


Click on the sky area with the eyedropper.

Adjust the selection criteria to get the whole sky area selected.

Hit the highlight button next to the screen to get the whole image and adjust the hue until nice and blue.



Check the rest of the clip and tweak the color as necessary.

Click the deliver button on the bottom and add the clip to the render queue and click the render button and that s it.

[Stan Jelavic]

Here is the daVinci output file.
https://drive.google.com/file/d/1K2o...ew?usp=sharing

[Stan Jelavic]

The sharpness in the above video is not optimal. The camera focus was a bit off. I readjusted the focus and took a test shot and then compared it against the BUC02. It is definitely sharper.

[David Brown]

I wasn't going to say anything remembering your real focus here is on the color!!!

Raw files means your capturing single frames! What is your shutter speed? At 2fps it must be faster than 1/30.

Is there any simple way of slowing down the Wolverine?

I just received the Winait version of the Wolverine that I plan to use for this project. It was heavily used by the seller for the last 2 years, and sold for parts. His descriptions of the problem sounded like electronics failure. I'm not sure.

On first inspection when powering up there was at times a "grinding gear" sound as noted by "Marco" as well as strange claw behavior. Film advance would stop for 2 audible cycles and then continue and repeat. I'm going to need to dismantle the claw cam and investigate.

The general problem with not recording or the machine not switching off, is I think, cause by a missing cue from the claw sensor. When pressing the "OK" button (on this machine that is how to begin recording) a message appears "Do not move the film". The claw begins moving the film, but the take up reel does not turn. That never changes until the plug is pulled. I found that just touching the moving cam wheel with a screwdriver caused the capture to begin recording. And it does record. There is crazy jitter. Left/right and up/down as seen on the tiny screen and on the recording. I wonder if some of the jitter is actually electronic? Watching the film in the gate, I cannot see any erratic movement side to side or up and down. Might be too small to see. The camera position is just an electronic zoom. Could the coordinates be dancing? just a thought.

My guess is the claw sensor is not always working, possibly from a mechanical problem. Will look into that. For now the mechanical noise has stopped with the back off.

[Stan Jelavic]

I have not played with the color extensively in daVinci but seems like it is possible to get it right. You need several nodes. The image comes out really sharp for a single shot but at 2 FPS the camera shake degrades it somewhat. Comparable to the board cameras and 16mm lens. The scan would have to be run slower than 2FPS. The only way of doing that is to use the Hawkeye board. I cannot think of any other way. The e-shutter cams should not have this issue.
I agree, you issues may be related to the optical coupled trigger. Maybe the cam mechanism is worn out and the clay is not engaging the optocoupler properly.
Wolverine Data sell some of the cam components as well. Once you figure out what is wrong, take a picture of the part and send them and ask for a quote. Once I got the cam spring from the for $8, shipping included.

[David Brown]

Hey Stan

https://www.amazon.com/photos/shared...VchY4CXBM8IGfM

I should have posted this with #56. The entire width of the film is not shown, but stills could be measured in IC measure. This amount of movement seems impossible. When 8mm is poorly split in half from the original 16mm width, errors can happen leaving the width over 8mm or less. In Marcos examples The right hand margin grows and shrinks while the image remains stable.

Here the image and sprockets all move laterally together. So either the entire gate assembly is moving, or the camera. The physical camera can't move (I don't think it can) so is the cropped image position moving when saved?

Just a thought, I won't be using the Winait's camera later.

For a single image, or even 100 it can be easier to process in Photoshop or similar still editor where you are more comfortable. I find video filters have different names or results. Photoshop can create a LUTs (just a preset) from a processed image that can be applied to a video timeline.

Good to know about the parts!!!

[Stan Jelavic]

The top of the frame is the shadow of the gate I believe. That looks stable, so I don't think that gate is moving. The camera sensor is soldered to the board. Perhaps the board is loose? Is the S8/R8 switch solid? Check the cables from the S8/R8 switch and and the sync cable and make sure they are not loose.
Here is the side-by-side of the Mark2 capture compared against BUC02.
https://drive.google.com/file/d/1WiG...ew?usp=sharing
Mark2 has clarity and a bit better resolution when compared visually. Mark2 original resolution is much larger pixel wise. The colors still need more work and a LUT would really help. Probably figure out the LUT in Photoshop as you suggested and then plug it in to daVinci. I use filters and sometimes there is an overlap with unwanted results. Example - the hills color at the end of the clip.

[David Brown]

I think I have the answer to the unstable images. THE FILM IS MOVING in the gate. There is only 1 pressure point, above the claw. I bet all these Wolverine variants are made in the same factory. The lid on the gate should be typical. That is one thing on my list to replace.


The rest has been dirt embedded in the grease. I don't know how the sensors worked at all, if they're optical. Everything is clean now and will see what happens.

Clearly the BUC02 wins in your side by side, though I do prefer the rich colors and contrast of the Cannon. Trying to ignore the focus problem is tough! I found WDR is a BIG factor when comparing my first attempts. Plus the BUC02 has a solid lens mount.

It's tough to mount a long lens and support from the camera's mount. If you move forward beyond this test that probably means a new camera and lens. That's $$$.