Engineering Randomness
@EERandomness
Followers
2K
Following
5K
Media
530
Statuses
6K
All Things Engineering. Electrical, Mechanical, Software, Firmware, AI, Security and everything in between. Specialize in custom HW/FW/SW for motor control
Joined April 2022
I am 100% convinced that Gemini's new image model (nano-banana) is actually a video model producing single frames. It is a world model. It can rotate objects with near perfect accuracy. This image in particular confuses every other model I tried due to the reflections.
108
236
7K
RT @ns123abc: NVIDIA HAS TAPED OUT 6 RUBIN CHIPS AT TSMC. - New CPU (Vera - 88 ARM cores with SMT).- New GPU variants (R100 with HBM4).- Sc….
0
12
0
YouTube is definitely testing something today. The volume won't go all the way up and a new UI keeps showing up
0
0
1
My wife is a Veterinarian. Whenever I have a bad day, I go read the latest Google reviews about her clinic. It never fails to remind me how amazing she is and how lucky I am to have her in my life!.
0
0
5
This plot is absolutely mind blowing. How can we explain this in a way that does not require an incredibly deep understanding of physics to realize just how crazy it is?. @eevblog any ideas?. It would also be fun to know where the spikes at 10Hz and 15Hz come from?.
We tested our controllers in hardware on the real LIGO system. Our results show that Deep Loop Shaping:.🔹controls noise up to 30-100 times better than existing controllers. 🔹can eliminate the most unstable, difficult feedback loop as a meaningful source of noise on LIGO for
0
0
1
I wonder which event/meeting in history had the highest total sum net worth of the people in attendance?. If Musk and Bezos had attended, this could have been it?.
BREAKING AI 🤖: . Trump is hosting tech leaders for a dinner at the White House. Guests include Meta CEO Mark Zuckerberg, Apple CEO Tim Cook, Microsoft co-founder Bill Gates and OpenAI CEO Sam Altman. Where is Elon Musk?
0
0
0
DON'T do this. This is hardware being used to bypass all the permissions on your phone. Those permissions are there for a reason, it basically gives their app the ability to listen at any time (mic) or send key presses (keyboard emulation) to ANY app.
Introducing AI Key, a small device that lets AI control your entire phone. just plug it in and ask it to complete a task. pre-order now.
1
1
16
The model loves to mesh things, and is incredibly accurate at it. If you tell it to "generate a wireframe mesh".It can then use the mesh to help it realistically understand how to "pile" parts
0
0
0
I gave nano-banana a single capacitor image and a few prompts to see if it could be used generate 3D data for recognizing a "pile" of PCB parts. The first image is real, all the others are generated. The idea is to use the generations to guide a robot. Not quite, but close!
1
1
11
CAM software will be among the first to be completely replaced by AI.
0
0
4
I worked on ASR (Automatic Speech Recognition) for a few years and it was shocking to see how people interacted with it. When it didn't understand them, they would speak louder, or slower, but almost never considered that the TV was on in the background or their car was loud.
0
0
0
I think the big reason AI labs are pushing voice interfaces is because it tricks your brain into forgetting you are talking to AI. When people speak to AI, they will ask it things they never would have considered if they were typing on a computer. Speech fundamentally allows.
Voice is an interface, and I feel we’re going to see voice explode. @OpenAI gpt-realtime is out and allows for native speech-to-speech and precise tool calling in real time. It can also take image inputs and place SIP phone calls - good for a quick triage before human handoff.
1
0
3
Very cool. @jimbelosic Any chance of SendCutSend carrying silicon steel (electrical steel) so we can order laser cut stators and rotors?.
Goodbye sore wrists, hello smooth torque. Meet our in-house BLDC winder.
1
0
2
Batch generating w. Nano Banana + Weavy. No prompt consistent product workflow. Input: 1 Image. Output: 20 images. (Full Weavy Workflow in thread). PROCESS:.01. Upload base image. 02. Create varied image angles. 03. Create new prompts per angle.04. Generate images.05.
0
0
2
Another example of the model understanding 3D. With some fine-tuning this could be incredibly useful.
3
4
56
@MacroWorld21 @Sim89776996 Here is another version and the original image if you want to try it yourself
0
0
12
This is what the part actually looks like for those who are curious
3
0
139
This is also a very interesting test I give image models. Most cannot generate a blank white (or any color) image because it would allow detection of their hidden watermarks. Gemini is no exception, it absolutely refuses to just generate a blank image.
16
8
491
By the way, this would be a hard test for most engineers. In particular, the screw plating requires use of subtle hints other than just the color/appearance. The screw in the middle is a curveball 🤣.
1
0
2
Not perfect, but amazingly close. I suspect if I told it to segment each part of the image first it would pass this test. (Gemini/nano-bannana). The reduction of resolution when feeding images into models still seems to be a bottleneck for CV applications like this
1
0
8