Update: check out my latest video - the most realistic simulated movie to-date of the supermassive black hole in the centre of our own Milky Way galaxy! [https://youtu.be/wU3vRKczamc]

On smartphones this video is best viewed using the youtube app. Try google cardboard on your smartphone with a VR headset for the most immersive experience (smartphone with gyro sensor required). For desktop computers try viewing in Google Chrome browser or Firefox.

If you're in a hurry, jump forward to 2:00 where it gets really interesting.

Science note: this is a real physics calculation (not simulation or rendering) I did of what we would actually see if we were unfortunate enough to fall into a black hole, from far away all the way up to the event horizon! Each frame is rendered at a resolution of 8K and the video plays at 60fps. Calculations are performed using my own general relativistic ray tracing and radiative transfer computer code, 'BHOSS' (Younsi et al. 2017), i.e., solving the equations of motion of light/photons (null geodesics) for a given spacetime, in this case a spinning black hole (Kerr).

For now there are no Doppler or gravitational redshifting effects for the sake of clarity. I've omitted an accretion disk and proper radiative transport of light as it distracts from the gravitational lensing of the starlight and the black hole's shadow. Including an accretion disk or even a torus calculated from a proper general relativistic magnetohydrodynamical computer simulation of gas falling onto a black hole is also possible and I may do this in the future.

In this movie the black hole is spinning rapidly (almost at the maximum possible rate). The starfield is taken from real observational data. The movie starts one thousand gravitational radii away from the black hole and ends at the event horizon, where eventually all light focuses into a single point and vanishes. Try looking around as you approach, or you'll miss it!

I made this VR movie to promote the "Einstein Inside" exhibition touring Germany, where it was first shown in November and December at the Goethe University of Frankfurt. Copyright: Ziri Younsi.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Movie produced from calculations utilising my general-relativistic radiation transfer code BHOSS, e.g., Ziri Younsi et al. 2012, 2016, 2020:
https://ui.adsabs.harvard.edu/....abs/2012A%26A...545A
https://ui.adsabs.harvard.edu/....abs/2016PhRvD..94h40
https://ui.adsabs.harvard.edu/....abs/2020IAUS..342...

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

This video covers the additional work and considerations you need to think about once you have a deep neural network that can classify your data. We need to consider that the trained network is usually part of a larger system and it needs to be incorporated into that design. We also want to have some amount of confidence that the model will work on unseen data and that it’s going to interact as expected with the other system components. Ultimately, we also want to deploy it onto a target device which requires certain performance characteristics.

Check out these other links:

• MATLAB Deep learning examples: https://bit.ly/3deCj40​
• 5 Reasons to use MATLAB for deep learning: https://bit.ly/2QlbNNc
• Getting Started with Deep Network Designer: https://bit.ly/2Qof12l
• Quantization of Deep Neural Networks: https://bit.ly/3daOj6u
• Multi-Loop PI Control of a Robotic Arm: https://bit.ly/3sevqDS
• Deploying Deep Neural Networks to GPUs and CPUs using MATLAB Coder and GPU Coder: https://bit.ly/3tfhEST
• Deep Learning HDL Toolbox: https://bit.ly/3tl88xF

--------------------------------------------------------------------------------------------------------
Get a free product trial: https://goo.gl/ZHFb5u
Learn more about MATLAB: https://goo.gl/8QV7ZZ
Learn more about Simulink: https://goo.gl/nqnbLe
See what's new in MATLAB and Simulink: https://goo.gl/pgGtod

© 2021 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc.
See www.mathworks.com/trademarks for a list of additional trademarks. Other product or brand names may be trademarks or registered trademarks of their respective holders.