Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Hans_Punk

Pages: [1] 2
Would you mind talking about the turntable a little bit?  I've been considering building something like that of my own but I have very little experience in the field.

I've currently been using the motor parts of my 3 axis time lapse rig from The motors can easily be adapted to make custom rigs or turntables, whilst using their NMX controller for the stepper motors. The advantage of using stepper motors for sliders or turntable use, is that you can have incredible fine control of position and speed. Using the controller, you can set repeatable moves/rotations at variable speed...or fire the camera - then move the motors (known as SMS - shoot move shoot), enabling fluid motion of moves even if your camera is shooting long exposure stills. The motors will only move to the next position increment after the camera has taken the exposure, and the file has been written to the card, just like you would configure for a moving camera move in a time lapse.

A good lower budget solution is to make your own small turntable with stepper motor drive with an Arduino type driver, or a simple DC geared motor drive. If using a DC geared motor for macro work , it is almost always best to use a 0.5rpm-3rpm motor with a PWM controller to help control the speed to allow ultra slow moves when needed. The PWM controller acts as a speed controller by using pulse modulation to help the motor to not stall under load at lower voltage. For closeup macro work, it is amazing how slow you might need a motor to turn....often hard to detect movement with naked eye, but when magnified at extreme macro it can easily be seen.

The best DC motor solution I found was to use a 30:1 gearbox combined with either a 0.5rpm or faster speed motor, as shown here:

Here are some example clips of abstract macro subjects filmed using Full Res Pictures + 3.5k crop mode on 5DmkIII.
Using 50 & 75mm lenses with +5 achromatic diopter with macro extension tubes to get some very tight shots - sometimes no bigger than a 10mm section of fluid inside a petri dish!
Some shots had a custom motorised turntable to move either the subject or camera a few millimetres at the desired rotational or tracking speed for the appropriate frame rate or scale of subject.
I'm still experimenting with crop mode for macro use, it is a lot of fun and I am really loving the possibilities it opens up.

Many thanks to A1ex and all the other devs who make this possible, and continue to improve the ML experience.

Reverse Engineering / Re: UHS-I / SD cards investigation
« on: April 06, 2018, 04:17:29 PM »
Ouch, that's not good. Can you try the card on a Linux PC (possibly an Ubuntu Live CD/USB) and show the last lines from "dmesg" after inserting it?

Those huge numbers sound like an uncaught error in the benchmark code.

Going to order one of these to check the signals, to make sure the controller is not somehow switching back to 3.3V while the card is in UHS mode (1.8V). Meanwhile, found out what driver strength means, and reverted the module to the previous changeset, just in case.

Do not try on expensive cards until we figure out what happened!

Cheers a1ex!

Unfortunatly I am unable to test on Linux machine...or get the SD card to register as showing up as a device at all. Shows all the signs of being completely dead (tried numerous device detection methods on PC).
It would make sense that the SD card could have been subjected to over-voltage in UHS mode, as the the card appears to be totally non-responsive (cooked) - not even 'recognised' as a non-recognised device to attempt a repair or recovery.

It seemed that running the ML benchmark test after running is what somehow triggered an error - maybe resulting in voltage change that cooked or 'knocked-out' the card when in UHS mode?
Before running the ML Card Benchmark test, I was successfully recording raw video at around the improved 45MB/s speeds (from 20MB/s original) which falls into similar results others have recently reported on Sandisk Extreme pro cards.

I purposely bought the SD card at full price from the top rated seller on Amazon - just in case something like this happened. Not a foolproof way to ensure a non-counterfeit product, but more reliable than some other online least an easy refund should be possible :)

Here is a video of the benchmark speeds on second attempt to check the card after clean re-install - It was obvious something was wrong, but part of me really wished it was a real stable write speed from the SD card that could be harnessed by ML
I can dream can't I? :)

RIP Sandisk

Reverse Engineering / Re: UHS-I / SD cards investigation
« on: April 06, 2018, 02:43:42 AM »

Benchmark from brand new SanDisk Extreme Pro 64GB 95 MB/s card - 3849.6 MB/s Write, 2836.5 MB/s Read!
Benchmark was done in a blink of an eye!!!

(After was run in photo mode with only Benchmark and loaded within 5D3 fw123 "4K raw video recording; lossless compression" version of ML).


Tried a test recording out of curiosity but was unable to...the camera displayed 'Card Full' message and refused to initiate recording.

After quickly realising that something was not right (or probably healthy for the card), I tried a clean install procedure (format SD card to Exfat, clean install of Crop 4k 123 5D3 build with in module root). As before, I was getting solid 20MB/s R&W before overclock...then up to 45MB/s with overclock but with apparent errors (like a couple have had so far with this card type):

SDR104 @ 120MHz: D0 D0 r:err [SAFE] D0  [BACK] D0
until the end...
SDR104 @ 160MHz: D3 D3 r:err [SAFE] D0  [BACK] D3

I was able to record at approx 45MB/s speed using the improved overclock speed.

Did another ML card benchmark...and R&W speeds are same as above screenshot. Tried a test recording again> 'Card Full' message again> Restarted camera> camera did not turn on.
SD card now appears to be dead as a dodo. Cannot get it to be discovered on win PC through disk management.

Removed SD card, camera turned on and booted fine with my slower Sandisk SD card with ML (all back to normal).
Foolishly I should have quit earlier to recover the logs from the SD card, but alas that Sandisk appears to be toast, hence the lack of precise data - just my anecdotal report.

So, the Sandisk extreme pro 64 gb 95MB/s card might not be a wise choice right now (for some) Unless I overlooked instructions from the thread about not running ML benchmark speed test after SD overclock?

SD Card was new item from Amazon reputable seller - 2017 version of the card according to the box.

I'm happy to take one for the team on such an early offshoot discovery (and potentially look stupid)...but figured it best to report in case it is of any information of value.

Just a confirmation from my opening post that there does not seem to be a solution that works. I’ve followed the thread so far and tested pretty much what everyone else here has tried (with variations) and have concluded identical results. Seems there must be a conflict in protocol when utilising usb as an external controller, in most cases triggered when initiating/using ML modules (especially MLV and Raw video modes).

It would be an amazing feature to be able to control the camera via wireless usb input solutions. Giving the 5D3 wireless, touchscreen control with a liveview feed to phone or tablet is a wonderful tool for remote triggering and monitoring.

Does it help if you disable lossless compression?

I believe I tried that with no success. I will try again tomorrow evening when I get back home - hopefully with some more concise findings to report.


Thanks Wayne H...I’m kind of glad I’m not the only one.

I can only confirm the problematic issues are with the Cam-Fi setup, with iPad, Android and iPhone all tested as app controllers (with same result). I do not have any other wireless usb or cable Tethered apps to test with at present. I also have not tried any other builds other than the 10th Oct 123 4K crop rec 5D3 build..since I’m currently away this weekend...I was curious to raise this issue in case others (more tech savvy than me would be able to chime in if they’d experienced similar control issue with a Tethered usb or wireless to usb setup like the Cam-Fi).
I too can confirm the same error 70, and I also tried various attempts with camera Live-view settings to see if I could trick the camera to stay online without errors (as it seems to do when in h264 mode - when no Raw video modules loaded). I even tried loading the simultaneous h264 proxy recording to see if that would fool it - sadly it did not. Curiously, throughout all errors on camera LCD - the live view feed on the app seems to remain live and uneffected. It is the control communication to the camera that seems to be the issue.

It seems that once the live view signal from active ML raw module is detected, then starts a chain of errors that causes disconnection of control and ultimately an error 70 situation. It is possible to trigger a start record command, but then the errors and freezing of the app/ camera make it unable to send a stop command. The camera can however stop recording if manually pressing stop record on back of camera - which at least gives me some glimmer of hope that the signal is in there somewhere, but being frozen out  as the avalanche of errors are initiated.

The Cam-Fi device is a really neat little thing, I can’t imagine it is too different from other wireless tether Wi-Fi app controllers. I assume it is something to do with the usb being the gateway of command AND display that for some reason is not liking the live view signal from the camera once raw rec module/s are loaded? The actual live view feed seems to be active/live and present in the app during the errors, but it is these overlayed errors on the camera LCD  that ultimately freeze the operation of usb control after record command has been initiated (or sometimes before upon raw module activation).

Apologies for a non formal error report, but it is tricky to explain without visually demonstrating.

Cheers A1ex

I recently bought a Cam-Fi wireless camera controller - and when in regular canon mode, not running ML...all is really great, the video live-view monitoring alone is superb.

As I half-expected, once trying the setup when running ML, a bunch of buffer errors prevent the communication to be made or sustained for anything other than a 'start' signal to the camera to take one still, or start a video recording...then the app and camera just freeze with a load or errors and will not play ball.

To be clear, connection is lost after trying to trigger recording when camera is set to MLV or ML raw module is active....h264 recording and monitoring works fine.
I'm running build 123 Oct 10th on 5D3.

Here is the product details of what I'm very keen to get working whilst running ML:

Here are a couple of error examples that happens when the Cam-Fi attempts to control the camera through the usb input (I have no clue what the specific problem is):

Before embarrassing myself any further by revealing my total lack of knowledge all things to do with interpreting code and such there an easy explanation as to why the communication to the camera control is unable to be achieved whilst running ML? - I suspects this issue would also happen if the camera was tethered physically?

All I want is to access live view and start/stop ML recording using the Cam-Fi wireless is a very cool little device for wireless monitoring of live-view and for full camera control when in standard canon mode. It just seems to only currently work for horrible .h264 canon video :(

Any advice welcome (including me being told it is an impossible dream)...any specific info I can submit to get closer to getting a dev to work out, let me know.


Share Your Videos / Re: Vietnam backpacking | 10 bit slowmo | 5DIII
« on: April 11, 2017, 01:00:26 AM »
Handy Seamless Transitions | Pack & Script | After Efects Project Files - Videohive template...

Great work @Savy!

Ah fault for not acknowledging that info. I should have noticed the custom mlv_dump disclaimer.

Cheers A1ex / Danne & GutterPump for reply x

Amazing work @a1ex as always!

I'm using MLV_Dump for the newest crop_rec module with higher resolutions (4K, 1080p48 etc) build - And I'm getting vertical stripes on the .dng exports.
Is this normal for now?...or am I missing something simple? I have not used MLV_Dump before now (and I'm a code ignorant idiot)
Test shots were lower ISO 100-400, various resolutions tested from April 4th build - all exhibited some vertical stripes (some more noticeable at lower ISO)

Running this code on Win 7 works for me to export the .dngs's, I just don't know if vertical stripe correction is an option yet in this version of MLV_Dump...or I simply need to somehow amend the .cmd code to 'switch on' the correction in MLV_Dump?

@echo off
md "%~n1"
"Q:\NEW MLV\mlv_dump.exe" -o %~n1\%~n1_ --dng %~n1.MLV

Link to screengrab of most prominent vertical stripes:

.Dng Sample files:

This can actually be very useful. Some users have been asking about this "feature" of shooting an H.264 along with raw video. It is a solution for a quick on set confidence check playback. If it is frame accurate it can also be sent to editorial to start the edit while the raw files are being processed.

YES! - This would indeed make a wonderful feature. Playback is probably the major bug bear I've encountered on shoots when trying to convince others that a 5D3 ML setup is working...trying to get others to understand how amazing the slower/low res raw preview playback can be very unimpressive to people on set who simply just want to check the previous take in real-time playback as if shot regular H.264. Having the ability to know that a higher resolution raw master is being captured separately....would be totally awesome. Extra bonus being the H.264 would contain audio track for scratch sound to help sync external recorder etc.

A 100% accurate framing of H.264 playback (confidence proxy) would probably be the most useful feature for production use and editorial efficiency that I could think of right now.

For shooting 2x it makes way more sense to shoot in the 4:3 aspect from the ML camera options...then de-squeeze the live view using the anamorphic display feature.
Don't know what camera you have - but if using 5DmkIII, 1600x1200 is a great 4:3 resolution when using 2x lenses as it will deliver 3200x1200 pixels at 2:1 aspect when de-squeezed in post. If you want a 2.35:1 delivery, simply crop the sides a little to around 2820x1200 to get that aspect. Whatever camera you have - if it can shoot 4:3 or as close to that aspect as possible - that is the correct process to get the most resolution you can from shooting 2x anamorphic.

monitoring wide aspects are much better suited on an external monitor that can de-squeeze the image for you, rather than add more strain on the camera doing it for you on it's inadequate sized LCD .
Having a larger external monitor massively helps with judging focus when shooting anamorphic.

I guess the only ML add-on that could be of use is to have a zoom control or 'auto scale' that takes the wide anamorphic de-squeeze preview - and scales it to fit the live view screen as much as possible, rather than retaining it into its scaled square letterbox frame. At the moment, an external monitor with desqueeze and variable display scale values is the better solution.

I owe you not a beer...but a brewery!

Many thanks again a1ex, truly outstanding work as always.
I really appriciate the time and effort you and others put in to help this community.

@A1ex....huge improvement already, many thanks!

Hard to judge if it is still distracting when footage in motion, rather than in a still....but to me it looks almost 100% clean.
Don't want to steal your time too much - but if you have any joy pushing it further, that would be totally amazing and much appreciated.

RE: script, please be gentle with me in explaining how to implement...scripts and line commands are still confusing for my tiny brain, still trying to grasp basics.  ::)

@Danne Many thanks...first time I've ever used ffmpeg (errr...or typed any command line!)  :o
Even I understood that - thank you for helping.

Here are two short clip extracts, first in context of footage...second clip is start of take (against warm white background of studio)

@A1ex, sorry - my fault I added confusion with my previous low res jpg uploaded grab.
All footage is 1080p, 25fps. iso between 400-800, 1/50th shutter throughout.

Do these appear every time you record H.264 in crop mode?

Is this a full-sized image? In raw, the same pattern repeats every 8 columns and it's caused by nonuniform amplifier gains. Basically, divide the image in 8 sub-images (columns 1:8:end, 2:8:end and so on) and match their histograms. Looks like avisynth might be useful here:

Yes - to both questions.

I'll look into avisynth and extracting frames with ffmpeg - vcodec, but I'm a total idiot when it comes to what I'm doing, and how to correct with code.
Many thanks for reply A1ex :)

Did a shoot yesterday on the awesome 5D3 build 1.2.3 crop mode update from A1ex . h264 was requested by client (they needed 'easy' footage to edit)  - For some shots I utilised the 1:1 crop mode from the most recent update and noticed vertical stripes in the footage. Camera was iso 400-800 1/50 shutter and only noticed stripes after downloading footage from card. I've not encountered this issue when shooting raw with the same settings and crop mode - I wondered if anyone else had this issue?

Is there such a thing as post vertical stripe correction for .h264 files that does not require a programmers brain to implement?

Here is am example of the vertical stripes I've encountered on .h264 footage, when using the crop mode:


- 3x3 binning in 720p (unsquashed slow motion at 1920x664 in RAW, image cropped only vertically; preview and H.264 are squashed)
- 1x3 binning: read all lines, bin every 3 columns (extremely squashed image)
- 3x1 binning: bin every 3 lines, read all columns (doesn't work well, image has artifacts)

The last preset could be useful for anamorphic lenses, but unfortunately I couldn't get it work without image artifacts. It might be possible to fix it by postprocessing, or it might be possible to find some register values that fixes it, I don't know.

This is incredible, having a fix for 3x1 would be outstanding for those (like me and many others) who shoot anamorphic.

Thank you!

Works great, and can confirm crop mode works with external monitor (smallhd 501) on build 1.2.3 perfectly.
If sensor area remapping has been 'cracked' - could we ever see a APSC sub-mode crop option?

Having ability to use 'APSC only' or super 35 cine lenses without vignette on full frame without significant loss of quality would be truly outstanding.

p.s - I also noticed the April 1st screen flicker gag again  :D

General Development Discussion / Re: Higher grey/hacked preview frequency?
« on: February 13, 2016, 03:00:39 PM »
wow...i'd love to have this feature, ideally on 5D3.113 build (it is the ideal thing that would make shooting crop mode focus accurate with external monitor).
I followed the links and read the info for a couple of hours as to how to compile and merge to get it to work - but I really still don't follow how to implement...I am such a simpleton :(

Is there any chance of this accurate 3x crop realtime preview making it's way to the nightly builds?

Hardware and Accessories / Re: Has Anyone used any of SmallRig's products?
« on: September 15, 2015, 11:00:07 PM »
SmallRig products are very good for the money. I don't have that monitor bracket, but I have many other parts from them which are great.

I have an Edelkrone EVF/monitor bracket which is slightly more expensive but is also very good.

Feature Requests / Re: Chroma Key in LiveView
« on: May 19, 2015, 07:45:58 PM »
That would be a very nice feature, but probably be very intensive to run live.

An update to the 'Ghost Image' feature would be just as useful IMHO - being able to overlay ANY still frame from the card (not just most recent camera still) would massively help align greenscreen/composite/Timelapse shots by seeing the layout & perspective of the previously shot stills (navigable) or any still frame on the camera card. using the thumbwheel to adjust opacity mix of the overlay would be icing on the cake.

General Chat / Re: April Fools joke? WTF ML!!
« on: March 23, 2015, 09:11:49 PM »
'You're standing there trying to figure out what's wrong shot after shot, staring at your camera with a perplexed look on your face like an unprofessional dumbass, missing key action shots, burning studio time, having some bridal party getting more and more irritated, or whatever the scenario, and all because some ML developer thought it would funny.'

- I'd say that is have just described a sequence from Mr Bean.

Pages: [1] 2