Diving head first... Astrophotography!

Canon T4i with new 300 mm telephoto lens

We'll I've thrown myself into the deep end again. This time its astrophotography... Struggling to get a successful image stack of any kind for the past three days. I've been trying my hand @ doing image stacking using my Canon T4i DSLR with static tripod setup pointed at some bright stars in city limits with very low expectations. I must be a crazy person - running out into the neighborhood in the middle of the night with my camera...

I just wanted to go through the motions and see if I could get a stack even if its ugly! I've used standard telephoto lenses @135mm and @300mm zoom and long exposures (5-15 seconds per shot) and attempted to combine with deepskystacker (using both dark's and lights). I tried on vega, denab, and arcturas as my focal spots because they stood out in the sky.


I've made mistakes and gained a lesson from each:

Arcturas - short exposure (1s)

(1) Arcturas - super duper cloudy - probably wasn't worth shooting at all, because i could barely make it out with my own eyes, and camera only picked up like 5 stars at most. Did it anyway cause I need to get my hands dirty. (lots of 1 second exposures) Probably didn't expose long enough to pick up anything meaningful.

Vega - too long an exposure (30s)

(2) Vega - a little better of a night partially cloudy - whisps of clouds went across some of the photos, but in hindsight all my photos were exposed too long (30 seconds) and the stars streaked. I also forgot to manually set it off of auto-iso for pretty much all the data.
(3) Denab - Clear night - visibility index was 23% (my gf told me from some app - no idea what) Neighbors turning on/off lights, i ignored this, and tried to move where in the neighborhood lights weren't directly blaring at the lens. Looking at the pictures, it seems obvious that I wasn't focused right. This was the first night i transition from the 18-135mm lens to the 75-300mm lens, so lots more zoom, but I guess cause I ddin't zoom in on the screen i couldn't tell that it wasn't focused...

Deneb - Focus it stupid!

Oooof! This is a slog. Regardless of throwing these stacks of 100- 170 photos into deepskystacker, I can't get it to produce a single stacked image it can never figure out how to combine more than one image? Can't fault the software when you have crappy input. I guess it can't register the pictures cause it can't find the stars? And so the journey continues...



PS - I just bought this EQ6-R Mount from Agena AstroProducts. It was a tough choice between agena and High Point Scientific. Simple fiscal difference made it simple. Probably going to shoot with my camera and these standard lenses while i hunt for the right Celestron EdgeHD... Wish me luck!

Git tipsheet

Great GIT tip sheet here:
https://www.reddit.com/r/git/comments/bwwqr3/git_cheat_sheet/


The following are git commands that I found useful:

To create a new branch with your existing content (you're already linked to a remote origin)
git checkout -b newbranchname

To see difference between branches:
git diff gitlogid file_that_you_want_to_see_diff

To merge two branches:
git merge

To look at git log more carefully:
git log --oneline --graph --all

EASY Semantic Segmentation Tool!

This thing is great! Rapidly can create semantic segmentation. The only feature I'd like to add is a slider for adjusting the watershed / image cv thresholding algorithm to change its sensitivity. That said, it looks like incrementally applying watershed can help you iterate to the regions you want segmented anyway

Download Binaries

Github Repository

Config file to revise classes and colors.

How to move docker data directory to another location on Ubuntu

I ran out of space on my root drive as i downloaded some NGC docker containers on my linux box. I immediately needed to free some space up. With a 8GB docker image, something needed to move.

https://www.guguweb.com/2019/02/07/how-to-move-docker-data-directory-to-another-location-on-ubuntu/

This lets you move it to another drive - which is the reason I was doing it, my root directory ran out of space. Then I made a ubuntu install thumbdrive and booted ubuntu from the drive. This is sometimes called a "live linux boot disk".
(1) I shrunk my windows partition so there was room for the expansion using windows disk management.
(2) Then I used gparted to do two operations to expand the size of my root drive. Apparently I couldn't just expand the root partition, i had to do a "move left" operation first into the hard drive space I had freed up, then I did a right expansion on my root partition which worked! (I couldn't do it in one operation - gparted gives errors.)

Delightfully gparted comes with ubuntu!

DeepFake Test!

https://youtu.be/aKJpW0yPmO8

I used code from https://github.com/AliaksandrSiarohin/first-order-model to do this. Deepfake technology is here. Whether that is good or bad. I did test the fashion weights as well. I can see there are several applications possible with this... It took me an hour or two to get everything working. It's astounding where we are today.

Recording Screen in Ubuntu

Ubuntu Screenshots
"printscreen" button. It saves to picture folder by default
https://askubuntu.com/questions/346864/how-do-i-take-a-screenshot


Recording Desktop
(1) If you want to record without downloading anything on ubuntu:
https://tipsonubuntu.com/2018/06/09/record-gnome-desktop-ubuntu-18-04/

(2) Another good screen casting software (open source) and works effectively while playing games - is OBS. https://obsproject.com/download) is made for game streamers and lets you manage multiple video streams simultaneously.

Setting up Linux computer for ML (cuda drivers)

As I've setup several desktop computers with ubuntu for on-prem ML, sometimes installing certain nvidia drivers on ubuntu can cause the graphics driver to throw an error while booting and then will give you a black screen. Here's a setup of commands to fix it:

1. Reboot machine and go to recovery mode using grub (Advanced Ubuntu)

    a. Enable Networking

    b. Drop to Root

2. sudo nvidia-uninstall #To remove all  remaining nvidia drivers, to ensure clean install

3. sudo add-apt-repository ppa:graphics-drivers/ppa

4. sudo apt-get update

5. sudo apt install nvidia-418 #Could try a newer one, but this was the most recent driver that worked for ML for me on my 2080Ti card

6. Reboot - this should've fixed it.

7. nvidia-smi #will show you if cuda and gpu driver version

Update: Windows + Anaconda with tensorflow 2

While tensorflow 2.1 is out now, I just wanted tensorflow 2.0 in my conda environment, after some experimentation this simple set of commands does it:
conda activate tf-gpu-cuda9
conda install tensorflow

To install tensorflow 1.14 in a conda environment:
conda activate tf-old
conda install tensorflow-gpu

First Oculus Quest App!

After fighting with several ADB driver issues, reinstalling unity, restarting, unplugging, replugging, adb kill-server, adb start-server commands I finally figured out how to get my own unity projects loaded into this device. In unity, I used the default/built android SDK tools that can be enabled during the install of unity. I had it working then did a firmware upgrade on my quest and everything broke. Then I installed Android SDK (studio) app thinking this would help. This may have helped with some adb version errors (41 instead of 40) that were coming with unity. By installing that, and then updating my environment variable to include the folder for platform-tools that contains adb.exe it appears to have helped with the version issue.

In addition to this, the key to getting it to show up in the command adb devices (where there were no listed devices) but I could see it in windows and navigate files was to go into the oculus app on my phone and re-enable developer mode. I had already done this before but because of the oculus firmware update, I think it resets it back to off. I was pulling my hair out trying to fix something with adb drivers when this is all I likely needed to get it working.

I did some quick write up to control my player model and then realized there was already a pretty decent player controller script that came with the prefab so I threw away my code and am starting with theirs until I find a specific need to revamp it. (just wanted to walk around anyway). A few assets later and a quick load and I have a fun looking scene! I can't wait to show my GF!

Now I've built and deployed apps for; HoloLens, Quest, and Vive! Cool! Gotta keep creating.

Image extracting from Videos

Extracting Images
In order to extract images for photogrammetry and/or machine learning, I needed a variety of images that are very diverse in the scene, if you extract all images from videos there isn't much diversity of perspective. In order to do this I installed ffmpeg and took some OBS screencaptures from gameplay and went to work:

Here's the line that extracts images

ffmpeg -i lucio.mp4 -vf "select=not(mod(n\,100))" -vsync vfr -q:v 2 img_%03d.jpg

https://superuser.com/questions/391257/extracting-one-of-every-10-frames-in-a-video-using-vlc-or-ffmpeg

Labeling Images
Once I had all the images extracted, I wanted to use them to retrain a yolo object detector, I found and used this tool. As an aside I jotti virus scanned it too - it was clean, I used the windows version. It was super easy to use! I'll do a subsequent post if i build an interesting AI model.

https://github.com/developer0hye/Yolo_Label