elvith

joined 1 year ago
[–] elvith@feddit.de 5 points 9 months ago (1 children)

They basically did. I bet they just used an ORM in the backed and then pointed the API endpoint to the user entity without filtering the fields. This results in a dump of the user table (although row by row indexed by users instead of a full dump)

[–] elvith@feddit.de 3 points 9 months ago

I had one group project for an arts and media class, where I feared, that I'd be the one who'd have to do all the work alone. It got assigned on the one day when I was sick and when I came in the next day, every group and task was assigned and I was left to team up with that one guy that everyone didn't want in their team, as everyone knew he'd do nothing.

On that day I managed to plan out the project with him and assign all task (while silently trying to keep the scope in a way, that I should be able to do everything alone, if he failed his tasks). I knew he was a huge Star Trek fan, so keeping that as the theme for the project and making a small Star Trek "parody" helped to keep him on board. But after that first day? As expected - nothing. He didn't talk to me, didn't do any tasks, basically ignored me. So I made some small changes and had to complete the project alone.

In the end, I even did the presentation alone, as my team mate was missing some classes as usual. The teacher came to me, said he knew I did almost everything alone and was impressed by the scope of what I still managed to complete.

Grades in these classes worked like this: You get a score between 0 and 15 points. 15 is best, 0 is worst. You failed the class, when your average rating was lower than 5 points at the end. The teacher said, that this project would be 14 points for each of us, but *hint* *hint* that I was free to distribute the points between us, as I saw fitting for the worke done. I had to distribute all 28 points between us and *hint* *hint* if I were to assign more than 15 points to a person those were valid, but capped at 15 for the calculation of the final grade.

Guess who got 28 points for this project.

IMHO a fair way to grade group projects, as long as you're keeping an eye on the discussion on why/how to distribute points, so that it yields fair results.

[–] elvith@feddit.de 4 points 9 months ago (1 children)

You need tough guys to patrol on these streets

Prompt/WorkflowMade with Stable Diffusion XL in Fooocus.

Initial image was redered with these settings:

{
    "Prompt": "a gorilla (dressed as police:1.3), Rim Lighting, (empty street:1.5), neon signs, raining, digital art, sign that says NIGHT CLUB, cyberpunk city background",
    "Negative Prompt": "police in background",
    "Fooocus V2 Expansion": "a gorilla (dressed as police:1.3), Rim Lighting, (empty street:1.5), neon signs, raining, digital art, sign that says NIGHT CLUB, cyberpunk city background, intricate, elegant, highly detailed, sharp focus, perfect dynamic light, professional composition, dramatic, cinematic, breathtaking, new bright colors, inspired, designed, rich",
    "Styles": "['Futuristic Cyberpunk Cityscape', 'Fooocus V2']",
    "Performance": "Quality",
    "Resolution": "(1024, 1024)",
    "Sharpness": 2,
    "Guidance Scale": 4,
    "ADM Guidance": "(1.5, 0.8, 0.3)",
    "Base Model": "bluePencilXL_v050.safetensors",
    "Refiner Model": "None",
    "Refiner Switch": 0.5,
    "Sampler": "dpmpp_2m_sde_gpu",
    "Scheduler": "karras",
    "Seed": 3644486667727718124,
    "LoRA 1": "sd_xl_offset_example-lora_1.0.safetensors : 0.1",
    "LoRA 2": "texta.safetensors : 1.0",
    "Version": "v2.1.864"
}

Then I outpaint left/right twice and a third time for top and bottom with a sligtly changed prompt that asked for neon signs in kanji, as it would only generate variations of "police" or "club" on the signs otherwise:

"Prompt": "a gorilla (dressed as police:1.3), Rim Lighting, (empty street:1.5), neon signs in kanji, raining, digital art, sign that says NIGHT CLUB, cyberpunk city background"

[–] elvith@feddit.de 2 points 9 months ago

Depends on how much current the drive draws. If the m2 SSD needs more current than your regular thumb drive, your phone might not supply enough power and it stays unusable.

Besides that, as long a the enclosure uses the standard protocol and the filesystem is supported by Android, it should work

[–] elvith@feddit.de 5 points 9 months ago (1 children)

Ransomware in Windows:
You need to allow macros to read this job application

Ransomware in Linux:
You need to run chmod +x application.ods.sh to read this job application

[–] elvith@feddit.de 10 points 9 months ago (1 children)

Farts per stomach?

[–] elvith@feddit.de 1 points 9 months ago

I also really liked the very first image, that I got. But it wasn't quite what I had in mind

[–] elvith@feddit.de 4 points 9 months ago (2 children)

I saw "bard" and thought of the typical DnD setting in a tavern. But since were using generative AI, I thought should do "Google Bard"

Prompt / WorkflowI used Stable Diffusion running on my PC with Fooocus 2.1.864.

The first step was the pixel art bard on the CRT, which was generated using the following prompt/settings:

{
  "Prompt": "a pixel art bard with a lute and a chat window displayed on an 90s style CRT PC monitor, the bard is in the colors yellow, green, red and blue",
  "Negative Prompt": "",
  "Fooocus V2 Expansion": "",
  "Styles": "['SAI Pixel Art']",
  "Performance": "Quality",
  "Resolution": "(1152, 896)",
  "Sharpness": 2,
  "Guidance Scale": 4,
  "ADM Guidance": "(1.5, 0.8, 0.3)",
  "Base Model": "juggernautXL_v8Rundiffusion.safetensors",
  "Refiner Model": "None",
  "Refiner Switch": 0.5,
  "Sampler": "dpmpp_2m_sde_gpu",
  "Scheduler": "karras",
  "Seed": 7611562024347322673,
  "LoRA 1": "sd_xl_offset_example-lora_1.0.safetensors : 0.1",
  "Version": "v2.1.864"
  }

Second step was loading the image as an input and using inpaint -> modify content with the empty area in the top right marked and the following settings (negative prompt, as it somehow really likes to render pixel art cats?):

{
  "Prompt": "a speech bubble containing the text google bard",
  "Negative Prompt": "cat",
  "Fooocus V2 Expansion": "",
  "Styles": "['SAI Pixel Art']",
  "Performance": "Quality",
  "Resolution": "(1024, 1024)",
  "Sharpness": 2,
  "Guidance Scale": 4,
  "ADM Guidance": "(1.5, 0.8, 0.3)",
  "Base Model": "juggernautXL_v8Rundiffusion.safetensors",
  "Refiner Model": "None",
  "Refiner Switch": 0.5,
  "Sampler": "dpmpp_2m_sde_gpu",
  "Scheduler": "karras",
  "Seed": 8793801103451750926,
  "LoRA 1": "sd_xl_offset_example-lora_1.0.safetensors : 0.1",
  "Version": "v2.1.864"
  }

[–] elvith@feddit.de 3 points 9 months ago

you forgot a word. To summon the bot, mention it followed by "draw for me "

[–] elvith@feddit.de 2 points 9 months ago

I use Voice audiobook player, that can do that, too. But when I switch devices, ... it's easier to pick up where I left, if it's at least separated by chapters (or as some MP3 CDs do every 3-5 minutes a new track).

Also I do sometimes buy mp3 audiobooks for a blind friend who prefers to listen to them on a CD player (buttons can be felt and its easier to use than a touch screen). But a single, several hours long mp3 is bad in this scenario. And as i didnt find a tool to split them easily, Audible exclusives were out of the question...

[–] elvith@feddit.de 2 points 9 months ago

Thanks, I'll try it

view more: ‹ prev next ›