This year I’m excited to present our work on generative adversarial networks for synthesizing solar resource and PV production data. Come check out our modeling and validation approach!
“Synthesis of multi-year PV production data using generative adversarial networks” Gregory M. Kimball, Camille M. Pauchet, Rasoul Ghadami, Alberto Fonts Zaragoza
Presented at: PVSC48 https://ieee-pvsc.org/PVSC48/ 11B: PV Resource Modeling and Forecasting II Friday, June 25, 2021 9:00 AM – 11:00 AM ET
We went on a trip to Maui to celebrate springtime in paradise and this time Mavic Air 2 came along. Come check out scenes from Ka’anapali, the Upcountry and Iao Valley!
Video capture by DJI Mavic Air 2, editing in Kdenlive.
Machine learning methods for “super-resolution” or “upscaling” are more accessible than ever. I was looking to make some out-of-print playing cards for Arkham Horror LCG, but I needed to upscaled my reference images by a 4x to achieve good printing quality. After some research, I found the Keras implementation of upscaling at https://github.com/krasserm/super-resolution to be an excellent option.
Using the EDSR x4 baseline with sample trained weights gave excellent results from PNG input images. The left image shows an upscaled sample, and the right image shows the input data. The images processed from 300×419 pixels to 1310×1778 pixels with the addition of bleed areas to improve the printing results.
When using input PNG images, artwork, text, and symbols all show excellent upscaling behavior even as they interpolate 90% new information.
On the other hand, compressed JPG input images deliver lower quality and more prominent artifacts. The left images show an upscaled sample, and the right images show the input data. Although the input JPG images have the benefit of being 31-33 KB as compared to the input PNG images at 230-280 KB, the noise introduced by the compression limits the fidelity of the upscaling process. Good to know!
We spent an amazing day driving down in Bug Sur and exploring the Old Coast Rd. Come see the Bixby Creek Bridge, Big Sur River, and Molera Point!
Video capture by DJI Mavic Air 2, editing in Kdenlive.
Here’s a great highlight from the Midtown Dreamers on Jan 5: Midtown Dreamers cover of Tom Petty’s Mary Jane’s Last Dance!
Video capture by DJI Mavic Air 2, editing in Lightworks. Audio capture by Zoom H4n, mastered in Ableton Live.
A few weeks ago, I decided to build a custom wall mount for my PC! I’ve been wanting to ditch my old Dell computer case and also make the heat sinks easier to clean. The key components are a PCIe extender for the GPU with at least 30 cm cable length, and a low-profile CPU cooler like the MasterAir G100M. The build is designed to rest on two angled blocks mounted on studs 16″ apart.
Update 2021-07-31: This is working great! The PC runs quiet due to excellent air flow, doesn’t take up floor space, and stays clean due to its elevated position.
Finally the time has arrived to tear out our kitchen lighting and install LED can lights. I had a lot of fun laying out the design, and even more fun drilling into my ceiling. So far the new lights have turned out great – we love the custom color temperature setting and uniform lighting pattern.
I chose the Halo HLBSL 6 in. LED Kit for this build and it couldn’t have been easier to wire up. The 6″ hole saw made a lot of dust but staging the wires and the circuit worked great from the start. No problem!
One of the advantages of owning a thermal camera is that you can inspect HVAC systems whenever you like! The mini-split heat pumps I work with provide a great example and so far this one shows no significant degradation in efficiency.
The left image shows the view of the evaporator from inside the building at 76 °F ambient, with minimum temperature of 64 °F. The right image shows the view of the condenser from outside the building at 87 °F ambient, with a maximum temperature of 106 °F. Pretty sweet device!
After I completed my Mt Umunhum ride, I turned my sights on another prominent feature on the horizon in San Jose – the Lick Observatory at Mt Hamilton. For my first exploratory attempt I followed the recommendation to use Mt Hamilton Rd but I found that route to be too narrow and with too many cars. Also its a few miles out of the way from most of San Jose, which led me to focus on Quimby Rd instead. Quimby Rd is a steep climb with average of 8% grade and several +14% sections but sometimes it takes this kind of drama to keep me interested in a route. Eventually I worked up enough courage to try for the summit and it was a beautiful ride.
With the new recommendations to wear face masks meeting a shortage of masks, I decided to bust out my mini sewing machine and make some DIY masks. Luckily the internet is full of wisdom on this matter – Anjurisa’s video “How To Make A Fabric Face Mask At Home Easy” on YouTube offers an excellent design for beginners (https://www.youtube.com/watch?v=_6UcIDaxfqk). For my masks I used the same square-pleated design with around-the-head elastic (rather than around-the-ears) and wire nose bridge. The only material I was ready to commit to this project was tee-shirt jersey knit fabric so the result was comfortable even though the sewing was difficult!
Gregory Kimball 