This was posted 3 months 5 days ago, and might be an out-dated deal.

Related
  • expired

Leonardo.Ai (AI Image Generation Tool) - Free 150 Tokens Per Day @ Leonardo Ai

810
Leonardo.Ai (AI Image Generation Tool) - Free 150 Tokens Per Day @ Leonardo Ai
Go to Deal
Scrooge McDeal on 23/01/2024 - 08:46 leonardo.ai (2655 clicks)
Last edited 23/01/2024 - 12:08 by 1 other user

Good morning, spotted this at My Dealz

The AI image generation tool Leonardo.ai is offering free image generation with the in-house “Alchemy” mode this week only. This allows high-quality images to be generated.

Simply register with an email and get started straight away.

Only 150 free credits can be used per day. How many credits an image requires varies depending on the settings you choose.

According to My Dealz, commercial use is allowed, but it is important to note that on the Free plan, all generated images are publicly visible to everyone.

Computing Artificial Intelligence Image Generator

Related Stores

Leonardo Interactive
Leonardo Interactive

closed Comments

  • PainToad on 23/01/2024 - 08:50
    +9

    Palworld's devs approve this deal.

    • smartazz104 on 23/01/2024 - 18:11
      +1

      Game Freak should have a look, might make a better looking Pokemon game.

    • frankfu1122 on 23/01/2024 - 19:20

      savage!

  • raj76 on 23/01/2024 - 08:54

    Any free ai image manipulation apps. Like remove trees or poles from behind you

    • Jonzay on 23/01/2024 - 08:59

      Are you after "AI-powered object removal" or just standard object removal apps?

      • BlackMamba666 on 23/01/2024 - 09:29

        Whats the difference if both can remove objects.

        • Jonzay on 23/01/2024 - 10:06
          +8

          I like TouchRetouch on Android (icon looks like a bendy pencil around an X), it does a decent job.
          If you have a Google One subscription, there's also object removal built into Google Photos.

        • Munki on 23/01/2024 - 11:27
          +11

          AI powered one can remove objects and fill in shit using prompts.

          e.g. Remove my wife and put in Elizabeth Olsen.

      • raj76 on 23/01/2024 - 10:04

        Ai powered please

        • Bii on 23/01/2024 - 11:25

          Sorry, I'm a bot an AI.

    • Davious on 23/01/2024 - 17:37
      +1

      Learn to deepetch. 0.o

      • raj76 on 23/01/2024 - 17:42

        What's that

        • Davious on 23/01/2024 - 18:54

          A missplaced comment. Im incorrect. Its taki g out the background. My bad.

    • shadowangel on 23/01/2024 - 18:34

      use google colab and fooocus, you can remove anything.
      https://colab.research.google.com/github/lllyasviel/Fooocus/…

    • PAEz on 24/01/2024 - 14:53
      +1

      Have a look for any tool that has inpainting. It allows you to select an area and then replace it with what you have in a text input. Theres lots.
      Photopea added some of this stuff, havent used it in ages, worth a shot.

  • Jonzay on 23/01/2024 - 08:54
    +15

    I know this is a "freebie", but they obviously want to draw you in to either make you view ads or get you to convert to a paying customer. Never spend money on these services, there's no need to.

    Anyone with a gaming PC can run AI image generation models of all shapes and sizes locally for free. I used to use a GeForce 1070 for it, and it had no issues with generating literally tens of thousands of images a day.

    • CptnObvious on 23/01/2024 - 08:56
      +2

      Agreed. Most people don't bother with the installation though

    • Zitane on 23/01/2024 - 09:05
      +3

      Do you mind helping with some suggestions on what program to use for this?

      • Ezuku on 23/01/2024 - 09:14
        +15

        Stable diffusion, which is a free local version. There are a lot of guides out there. Way more customisable, you can train it yourself, etc. It is quite easy to set up and there's a lot of material online you can download.

        Just be aware that you need a really powerful pc unless you want to wait a while; a 1070 would be quite painful and really pushing it. It's also heavily limited by vram. A 2060 for example would take about 40 seconds with a 512x512 image, and about 2-3 minutes if you wanted to upsize it (essentially rerunning it through the process which increases the quality, which you do). Longer if you wanted to do fancy things. A 2060 wouldn't have enough vram to make custom trainings (eg loras) or higher resolution images.

        Not sure what the state of play is but amd is also extremely finicky and I'm not sure if it works fully. Cpu only is very slow.

        • Jonzay on 23/01/2024 - 09:23
          +2

          The 1070 was fine for pure image generation, which is what the vast majority of people will use it for. The mobile apps tend to make you wait a bit too, so there isn't much difference there. I ran it in massive batches and did other things while I waited.

          I have a 4070 now, which has quite a bit more pep.

          • Ezuku on 23/01/2024 - 09:35

            @Jonzay: Fair enough. To be honest my 2060 was painfully slow especially when you need to upsize images. It also choked on control net. I also upgraded to a laptop 4090 (so roughly 4070 ti) and it's just night and day, hard to imagine going back.

        • Zitane on 23/01/2024 - 15:37
          +1

          thanks my dude. didn't know this stuff was available to consumers

          • Ezuku on 23/01/2024 - 17:24

            @Zitane: It's actually very easy to use, if you're the slightest bit interested I recommend checking it out.

      • mstlx on 23/01/2024 - 09:19

        Yes please share

      • ju2au on 23/01/2024 - 10:26
        +9

        With Stable Diffusion, you can use a package manager like Stable Matrix to install everything with a couple of clicks. Much easier than the normal install process.
        https://github.com/LykosAI/StabilityMatrix

      • glide on 23/01/2024 - 14:30
        +4

        pinokio (https://pinokio.computer/) makes a lot of local LLMs and tools one-click install and use

        • Shaolinking on 24/01/2024 - 16:58

          this is interesting, thanks for sharing

      • Thaal Sinestro on 23/01/2024 - 17:49
        +1

        If people just want to upsize images, there is Cupscale (a bit older now, a bit buggy) or Chainner (works better than Cupscale, uses nodes so it take a little bit of trial and error to get it set up properly). There are various AI models you download, each of them is only good for certain types of images on produces subpar results on other types of images. If you use a Nvidia Gpu you also need to download PyTorch, which is 5 Gb uncompressed. You have an option of a portable install (all Python requirements in a programs directory), or installing Python standalone and stashing the dependenices in Pythons directory (better if you are using multiple AI programs).

        Personally, I mostly prefer Topaz Gigapixel but it isn't free. I haven't experimented with Stable Diffusion yet but I suspect it has some similarities with Cupscale and Chainner, but it is more powerful since it can generate new content rather than enhance existing content. Photoshop also has generative AI now, but you can just use it to remove things like watermarks or powerlines.

        • PAEz on 24/01/2024 - 15:08

          If you can handle Chainner (still mean to have a good look at this myself) then you should also look at ComfyUI and have even more options.
          The typical process is to upscale the image using what you use (same as in cupscale/chainner) and then it uses another thing to add the details. These "things" are components you can chain together just like in chainner.

          For those that would like to see what effect these different models he mentioned can do have a look here…..
          https://phhofm.github.io/upscale/introduction.html

          And for a list of up to date models look here…..
          https://openmodeldb.info

      • PAEz on 24/01/2024 - 15:40

        Another suggestion is Krita with the SD plugin…..
        https://github.com/Acly/krita-ai-diffusion
        ….check its release notes to see everything it can do, the main page isnt updated that often…..
        https://github.com/Acly/krita-ai-diffusion/releases
        …its nuts how good this is getting.

        This is made for artists more than those that want total control, but the advantages of being able to do it in a free paint app is great.
        For those that dont have the graphics card to handle this they have an image you can use on vast.ai, this is what I do. Its a little painful at times to start but its cheap compared to everything else. I typically get a 3080+ for 16c (US) an hour with 30gig storage, for a less powerful gpu you can go below 10c..

    • cmyk on 23/01/2024 - 09:05

      New to AI, do you mean something like Stable Diffusion when you say running it locally?

      • Jonzay on 23/01/2024 - 09:20
        +2

        Yes, I am referring to Stable Diffusion and the various models available for it. Unfortunately at the moment I don't have the time to link guides, but I can see a few other commenters have given links and extra context, which I appreciate.

    • CharcoaI on 23/01/2024 - 09:19
      +3

      For the people asking for more info:

      (I haven't watched/used this guide personally, but given the view count it should be helpful)

      https://www.youtube.com/watch?v=Po-ykkCLE6M

    • tessel on 23/01/2024 - 18:02

      Last I checked SD whilst good was nowhere near the quality you get from midjourney. However, most people will be happy with SD diy but the quality wasn't on MJ levels and as mentioned below diy can be slow.

  • Dolt2020 on 23/01/2024 - 09:56
    +6

    For those wanting to D.I.Y and do it locally on their PC, you are after a front end to operate on the open source Stable diffusion models.

    I use Automatic 1111 (mostly). Installation is a little finicky, but not too bad if you follow the instructions - auto 1111 Install guide

    Another - EasyDiffusion - if you want to get up and running easily, pick this - one installer for the front end and it will then go and fetch the Stable diffusion model and you can start -easy diffusion d/l and install

    • rakkie on 23/01/2024 - 13:28
      +3

      These days I would say Fooocus is probably the least intimidating if you are just starting out.

      And ComfyUI if you want an advanced setup.

  • tightm8 on 23/01/2024 - 10:41
    -1

    All these 'startups' are just the same thing you can replicate yourself, wrapped up as an 'innovative' company that is being funded by VC's today, only to start spamming you with ads in a year or two to recoup the investors money.

  • BarkGains on 23/01/2024 - 10:41

    How dare they steal the name of one of the best ninja turtles!

    • Ninternet on 23/01/2024 - 11:01
      +9

      one of the best

      He's definitely top four

      • Bii on 23/01/2024 - 11:27

        Yes but I feel they are all dumb lol

        • CptnObvious on 23/01/2024 - 11:58
          +1

          They managed to bag April. Or is she dumb also.

          • Munki on 23/01/2024 - 12:19
            +1

            @CptnObvious: Casey bags April. The turtles just watch - Sneako style. ( ͡° ͜ʖ ͡°)

            • CptnObvious on 23/01/2024 - 12:20

              @Munki: Hmm, not in the version I watched…those turtles be packing

  • ferguson on 23/01/2024 - 11:56
    +1

    For those who want an alternative to this one, Bing Image Creator (Microsoft) also provides free credits.

    https://www.bing.com/images/create

    • Munki on 23/01/2024 - 13:58

      Bing uses Dall-E 3 as well, which is meant to be one of the best ones.

    • Bii on 23/01/2024 - 18:54

      Is it just the same as Copilot?

      • BuyOrNot on 23/01/2024 - 23:36

        Yes sort of, Co Pilot and Dalle use Open AIs models.

    • lansalot on 24/01/2024 - 20:57

      I'm a long time Leonardo user and Bing beats it for image generation hands down. It's much easier to get the initial images I want with Bing / Dall-E 3 and then use the Canvas tool in Leonardo to do the rest.

      As with any AI image generator though, it's all about knowing how to write the prompts.

      Leonardo's strength is the tools for refining images you've already generated.

    • ShipShapeRC on 28/01/2024 - 20:57

      Thanks, yes the Bing one is waaaaaaay better. It actually makes a picture like I asked, while Leo did something similar but quite different to what I wanted.

  • dojomojo on 23/01/2024 - 12:38
    +1

    Can we use it to generate an image of what the average Ozbargain user looks like?

  • Dobe on 23/01/2024 - 12:43
    +1

    OK, I tried it, with a specific instruction concerning an Australia Day image I wanted.
    Using the same instruction, then went to the Microsoft Edge brower on Bing's inbuilt "Image Creator" from Designer (also free).
    The winner of the results is really clear, "Image Creator". It is superior in every way.

    • Munki on 23/01/2024 - 13:59
      +1

      Cause Bing uses Dall-E 3, which is supposedly one of the best AI image creators.

    • lansalot on 24/01/2024 - 21:01
      +1

      You're 100% correct about the initial image generation.

      I'd recommend now loading that image in to the Canvas Editor on Leonardo and playing around with things.

      Tip: When you've accepted a change in Canvas it creates that change as a new layer over the original image, you can ashtrays drag that layer to the side and keep going to get what you need.

  • LowRange on 23/01/2024 - 12:44

    Tested it, tried to make a website logo saying a specific word. Failed miserable and done misspellings everywhere. Maybe there's a setting to stop that which I couldn't find, but the images were nowhere near suitable anyway.

    • Ezuku on 23/01/2024 - 14:07
      +3

      Yeah, you can't really do that. AI is terrible on things that need to be in a specific order, specifically text and hands.

  • PlushToys on 23/01/2024 - 15:28
    +1

    Low res junk.

  • illogicalerror on 23/01/2024 - 16:26

    I had a bit of fun with the site. Like, I'm never going to load an AI model on my own machine and it's not really my thing, but I prompted some pretty cool looking images, was fun.

  • whyisave on 24/01/2024 - 01:10

    Meet Nightshade, the new tool allowing artists to ‘poison’ AI models with corrupted training data
    https://venturebeat.com/ai/meet-nightshade-the-new-tool-allo…

  • SlickMick on 24/01/2024 - 09:13

    IMO AI is at about the same level of maturity as speech-recognition was in the 90s. It doesn't quite work, the results are unreliable, but some find it fun to play with.

    A stock trading company generated an image of a bull and a bear, so I've tried a couple of these tools to see if I could produce something similar. Every time, it gets things wrong, like claws on the bull or horns on the bear.

    I asked a text-based tool to draw ascii animals. It would just generate a random animal, rarely the one I requested.

    It scares me that anyone would be putting any faith in AI results.

    • PAEz on 24/01/2024 - 14:49

      Except with speech recognition we had no control, you can with this.
      You prolly used something based on Stable DIffusion and yeah without some help itll usually produce mistakes.
      Help can be Lora's (the right lora can make all the difference), control nets (helps with placement, matching source and all sorts of things), upscaling (upscalers/detail adders can fix things often if used right). Theres also ways to fix mistakes such as inpainting, hand fixers, woteva.
      No SD wont always produce what you where after, but it can be coerced to do so.
      What you want is to try MidJourney or Dalle-3 (and preferably not Bing) those two can understand text input WAY better than SD and give you closer to what you ask for (be specific and detailed).

  • poweredbysugar on 24/01/2024 - 11:37

    The real time canvas is pretty cool. For those that want lots of customization options with a decent user interface and no setup I'd recommend this.

  • PAEz on 24/01/2024 - 17:10

    Here's another text/image to video app that seems to be free at the moment…..
    https://app.pixverse.ai/create/video
    ….can do some cool things.

Login or Join to leave a comment
Login Join