Invoke 5.3 introduces a new Select Object Tool and Flux Support for Global Reference Images

[@foulwin9719]

This is why I use Invoke, once you start seeing how easy and artist friendly the tools are it's very difficult to go back to other options.

[@zyang056]

The presentation styles are so different it's hilarious. Good job team!

[@slowmotionradio6015]

Man, the 5.x generation is making Invoke a real competitor in the art space. I've been following development since 2.something and 5.0+ is when Invoke finally became what I was hoping for. It's my daily driver now, and for the foreseeable future.

[@cristitanase6130]

Amazing! Now, if you can add a rig to them and make them change posture you will have something on your hands no one else has. Another idea: Add a Face Module where you can place an image, and a prompt, and then can be used for all generations. If you can pull this one you will move ahead everyone else in the AI graphic generating space.

[@invokeai]

Love these ideas!

[@hsoj9550]

Krita actually had the rig posing support via the ComfyUI plugin (been a bit since I played with Krita for SD, but I remember that). So this should 100% be doable for sure. I'd love to see that, as it would absolutely make Invoke go to another level if that can be added!

[@hsoj9550]

This is absolutely incredible! Well done, y'all!

[@Mr.SpicyIce]

Great work as always guys!

[@SonidosEnArmonia_1992]

this is great. considering trying this tool again. Tomorrow might be the day.

[@TVdosPatriotas]

Amazing work, dude!

[@Digildon]

Cool

[@koufdell]

thx meta for SAM2

[@rodrigop.9071]

I will make sure everyone knows what a wonder INVOKE AI is.

[@hhaaasshh]

love you, many thanks!!

[@trigor_ipseite]

This is fire!

[@echorises]

You guys are creating the next Photoshop (in the sense of usefulness, not the Adobe toxicity stuff)

[@mchammer5592]

Let’s see those edges! 😉

[@Comenta-san]

😯😯😯👍👍

[@build.aiagents]

PHENOMENAL

[@USBEN.]

Impressive.

[@cmds.learning7426]

amazing！

[@hellcore]

Its a fantastic tool. I am honestly very terrible in drawing. Thats why I never paint. This tool gives me the opportunity and creativity my hands cant do :) Thanks. One feedback - the flux implementation seems to be very aggressive to the hardware and consumes much more memory than other tools like Forge or ComfyUI. You guys working or looking into that?

[@dhanang]

Damn, It might be the time to get back to Invoke after a year.

[@havemoney]

You are probably familiar with A1111, can you record a video with explanations for people who have previously worked on A1111 ? Example, you are familiar with the Inpaint option. Here it is implemented as follows: - we show it.

[@echorises]

Why though? If it took you 10x time to figure out A1111, you will master Invoke in 1x time. It is a streamlined application, not a rolling-release experiment zone with no UI/UX designers like A1111.

[@DezorianGuy]

Are there tutorials apart from the official ones? I have problems following those ones as they don't feel like real life scenerios (architecture etc) and are not step by step. Also is there a regional prompter do define where a person or object should be placed?

[@quincy3d69]

some kind of way to group/organize the loras and models would be nice ! so that i can choose folder in my concepts drop down menu. with many loras and models it gets kind of messy and hard to find the right one. and little preview picture like in the model manager would be super nice. thx and keep up the create work !

[@LittlePixelTM]

Does this version allow use of Flux models on a Mac yet?

[@cantstopthefunk22]

Are there ways to incorporate this tool with another platform like ComfyUI? My thinking here is that if there was full integration, then we could use the layer functionality of Invoke to further aid in the generation process. For example, if we used Invoke to create isometric images of a futuristic gun, then used that input in Invoke to generate a reference image to be used in ComfyUI (for video generation, for example), we could use the the previously generated isometric views of the gun as reference images, and we would use the layer feature to "label" the elements of the image that we wanted to keep consistent between frames. All we would have to do in the prompt is to say "steampunk character holding "laser_gun1", and rather than have the AI generate a new design every frame, it would pull from the reference images of the gun to create consistent generations. This could be used for characters, environments, anything.

[@madeupname-oe5kp]

you could just drag the image into invoke. Just drag and drop into the gallery

[@kymatekk]

any advice on why some flux loras from civitai dont work with invoke? also some models, like the one called colossus on civitai, it gives me "server error" when prompting.

[@invokeai]

Right now, there’s a wide range of different formats being used by model trainers and fine-tuners across the ecosystem, and unfortunately, there isn’t any clear standardization. Model trainers are often not specifying the formats they use, which can cause issues when uploading models. You can see the formats we have supported here: support.invoke.ai/support/solutions/articles/151000199277-using-flux-models-on-invoke

[@SzcZ]

Would it be possible to install v5 alongside v4 in Windows? Or would there be clashes with xformers version or another component...

[@jonchambers8864]

I have a few different versions for testing and they seem to get along fine for me.

[@tekratek4077]

We want SD 3.5 Large support!

[@invokeai]

Very soon!

[@havemoney]

@@invokeai and medium ) Please read my comments

[@tekratek4077]

@@invokeai Nice to hear!

[@AlexxSalles]

I can't use this feature. I always receive an error message. I've tried in three different environments: Docker, a desktop installation with Windows 11, and an online VPS. I always get the same error.

[@DylanDValentine]

LET. THEM. COOK.

[@eydaimon]

sorry if this is a dumb question, but how do I load the image to work on?

[@madeupname-oe5kp]

Just drag it from your other app, into the invoke gallery.

[@rifz42]

it looks cool but the pricing is not clear, I can't afford the monthly subscription but it says get STARTED for free.. what does that mean? is it free if I use my own computer and nothing on your online service? or is it just a demo? thanks! you should say "free time limited demo" or something.

[@damiantriebl1747]

I need to do battlemap for dungeon and dragons, is possible with that?

[@LouisGedo]

Hi 👋

[@streetphone4619]

I tried the Flux from your starter models. My 8 GB 3070 Ti laptop system - ROG Strix - chokes and dies on it. Please add the most low VRAM friendly version you can find. Please.

[@vgf64-27]

Precisely why I'm trying to get myself a PC with a high-end card. I'm also stuck with an 8gb laptop.

[@invokeai]

We have the quantized versions included as well. If those don't run on your hardware, you'll need to upgrade your hardware. Flux is a very VRAM intensive model.

[@streetphone4619]

@@invokeai yeah that seems to be the case but I'm not going to be doing that so I guess I'll contact myself with stable diffusion 1.5, XL, and Pony.

[@streetphone4619]

@@invokeai Oh and thanks for the quick reply. I love your app.

[@AI_Creatives_Toolbox]

The select object feature seems amazing! But for some reason, the action just keeps looping endlessly and never really finished the selection. Has this happen to anyone?

[@AI_Creatives_Toolbox]

Never mind, one PC startup later and it works like a charm!

[@EH21UTB]

how does one update the local version?

[@invokeai]

Download the latest installer from Github, and select the same folder. It'll only install updates - your models and images will remain intact.

[@Asyouwere]

Trying to run Flux-dev-fp8 is extremely slow somehow, like it runs on the CPU instead of my 4090.

[@zabique]

u sure it's on CPU? how long for 30 steps?

[@omegablast2002]

i wish it had built in hires fixing....

[@bubaondesign]

CustomXLv7 Model link please!

[@invokeai]

civitai.com/models/719084

[@bubaondesign]

@@invokeai Thanks A lot! I love the new Invoke ;)

[@kcoppa]

Pop pop pop pop. segment

[@plainpixels]

I’m not going to say it’s not cool, I’ve been using the same thing for inpainting with sdxl for like 2 years, and it’s great. But it’s hardly invokeai magic. 2 years ago, it was 30 minutes of Gradio and segment anything. Today it’s probably 5 minutes of ChatGPT and segment anything 2.

[@GeoBook]

Invoke is a fantastic tool! But the guy in the video focuses so much on showing what can be done that he completely skips explaining how to actually do it. Clear step-by-step instructions would make this video way more helpful.

[@Didi_Ki_Kahanii]

Is it free?

[@invokeai]

The Community Edition is free: github.com/invoke-ai/InvokeAI/

[@aa-xn5hc]

I tried it. the selection SUCKS completely

[@jazt89]

Another update? I still haven't learned to use the last one properly and they've already changed it again, at least they make tutorials in Spanish please

[@zephilde]

This AI voice is less expressive as yours Kent! Don't make it the default please! At least try a voice clone AI ! ;)

[@quizblog]

Guys 49 USD per Month for this? Is this a joke? Like what is wrong with the world these days?

[@invokeai]

Community Edition is free. Scroll down on that pricing page and you'll see a link to download.

[@machineman268]

I paid zero for mine.