Olivio Sarikas, an AI Expert and passionate Artist, invites you to explore the exciting world of AI art. Join live streams to turn your creative visions into reality. Let's create together! 🎨🤖🚀
Olivio Sarikas is a professional Designer from Vienna Austria, a passionate Artist for over 26 years. He studied at the University for Fine Arts in Vienna and Munich and holds a Masters Degree in Fine Arts.
Community Guidelines: Inspire, Engage & Help: Use comments to share passion, inspire, and support AI art. Respect and Kindness: Treat all with respect, regardless of opinions or backgrounds. Constructive Criticism: Offer constructive feedback; avoid overly negative comments. No Harassment: Report and avoid harassment or bullying. No AI Haters: Spread love and passion; keep comments positive and inspiring.
I love how a lot of these sites are now calling them prompt refiners. 😂 I've been doing a prompt refiner on my site and for my people for over a year now. I still call it a prompt generator though. Just because it's got a lot more flexibility than just refining your request.
Also for those of you who are questioning why you would pay for flux, some people can't afford the hardware to run it decently. I think sites like this are great for the masses who just need to be able to generate stuff quickly and are willing to front a little dough for it.
An entire paid promotion video? I get why you are doing this but I think in the long term this harms the integrity of your otherwise excellent youtube channel.
It has the "Load VAE" node. What VAE should I use and where can I get it? I see there are some VAE in civitai (a lot less than regular models), but I am not sure what to use.
Do you know if we can make a 3d model out of an image on CGDream? Also thanks for promoting my tortoise riding pommeranian render which I submitted last night. Just barely saw it in your video 😅
They should really list the NSFW 'sensitive content' is not FREE and list it under the premium tier in their pricing page. Just wasted time trying out the free version for no reason.
You can make Flux have a negative prompt but it needs a CFG > 1 and it doubles the generation time and it makes it so slow, but if you are willing to wait it can make more realistic-looking images.
Is it the same group as CGTrader? I've used CGTrader a lot over the years for Blender models. One feature they could add is some kind of motion model to do AnimateDiff or Deforum.
Simply browsing or searching the gallery requires logging in. I would prefer to look more into what other people have generated using this service before setting up an account for yet another online service.
So I’ve been using ChDream for some time now. They recently made some updates to the platform: They changed their pricing platform in the middle of people’s plans and took away features and move them to higher tier options. They increase the pricing across the plans without discussing anything with their install base, There has been an open discussion on their discord about bad business practices and egregious monetization. They also charged people that had canceled their plans due to a “bug” In their credit card processing. Apparently, the models they’re using (Flux dev) is charging for more than other platforms And using your generations to finally tune their model. Just some food for thought.
I don't need the rest but the 3D model feature sounds interesting and the amount of credits for free is decent. Some sites don't give you anything and want a month's pay with no trial.
a geforce 1070 i found in my closet from my closet runs flux, and a 1080ti runs it rather well.. let alone people who have anything more recient. i imagine the only people its out of reach from, are people running computers as powerful as chrome books or people who slum it on cell phones.
What do you mean by can run it and rather well? Flux takes 2 min 40 seconds the q4 model for a 1152x896 image on my laptop with rtx 2060. I wouldn’t call that it runs finr. I would call it barely runnable like only run it if you are desperate for a better prompt understanding . In the whole time it takes to generate a flux image I could generate AND upscale an image by 1.5 in sdxl
flux will not do genitalia AT ALL, and the Loras that add that do it, but not too well, for being so novel. flux will work just fine on a GPU with insufficient ram (and the larger model-size sure shows its capabilities) , where ever the video card driver allows to use mainboard-ram and gpu-vram for cpu-Compute. the modern nvidea driver allows this by default, and also has a disable-this-option.
remote services. text2image is quite energy-inefficient on home computers, and much more efficient on larger gpu-server farms. "game as service" MAY just make sense here. Games like ai-rougelike and VersesOfEnchantment exist, where you may not want to wait even 2 seconds per image.
Open browser, add one tab and load workflow1.json, add another browser tab and load workflow2.json. Now we can run parallel projects within same browser no matter on edge or chrome. Am I correct?
"Upscale latent" before "VAE decode" is much better than upscale the output image with more detail. This lesson integrates Upscale and KSampler, one more step for beginner.
What are the best settings (AI model, Sharpen, Denoise, Fix compression, face recovery, gamma correction), if I want to upscale Midjourney images to 300 DPI and high quality to use it for Print on Demand Model (Printify)? Thank u for advice.
Nice video Olivio but have to say was a little disappointed as I was expecting you to go through all the options and give a brief explanation as to what they do etc. :)
With James Cameron joining the board of Stability AI, it's over. The age of AI creativity that's open for the masses is dead. Done. Finished. You might as well start thinking about some other kind of RU-vid channel or an entirely different hobby.
@@OlivioSarikas Thanks, I assumed that might be the case. Anyway, I've heard that it's not really reliable at the moment and results might vary. Interesting stuff nevertheless.