Tu passes des heures à chercher le bon prompt pour tes figurines ? J'ai fait un outil pour ça. http://promptbeast.fr PromptBeast est un générateur de prompts gratuit pensé pour les makers. Tu configures ton personnage en quelques clics, tu copies le prompt, tu le colles dans l'IA de ton choix (ChatGPT, Gemini, Midjourney, Leonardo...) ou directement dans Meshy Text-to-3D. Ce qu'il y a dedans : 🐾 Constructeur de personnages — Mammifères, oiseaux, créatures, gaming, anime, films, comics... Des centaines de combinaisons avec poses, accessoires, émotions et morphologies. 🌄 Mise en scène — Fonds studio, nature, néon, médiéval, japonais... Pour tes photos de figurines imprimées. 🦎 Flexi animals — Prompts optimisés pour les animaux articulés. 🏆 Concours — Un onglet dédié pour générer des prompts autour d'un thème de concours, avec angle artistique et style au choix. 📢 Publication — Génère directement ta description MakerWorld, Printables ou réseaux sociaux à partir de ton modèle. En français, anglais, ou les deux. ✨ Inspiration — Cards prêtes à utiliser et tweaker si tu manques d'idées. Gratuit, sans inscription, dans le navigateur.

We talk a lot about the democratisation of manufacturing. But spend time in this community, really spend time, and a harder truth becomes visible.The machines have never been better. Bambu Lab's A1 is practically a household appliance at this point. Self-calibrating, smartphone-connected, quiet, fast. The hardware barrier has been dismantled.And yet printers go idle. Filament spools go stale. Machines end up listed at half their purchase price.Because the barrier was never the machine. It was always the model. A 3D printer without a 3D model is an expensive paperweight. And creating a functional model, one that is dimensionally correct, structurally sound, ergonomically sensible, and actually printable, requires skills most people don't have and have no particular reason to acquire. The learning curve for CAD isn't steep. For most people, it's vertical.So users do what's rational: they come here, to MakerWorld. They look for something that already exists. When it doesn't exist in exactly the form they need, which is often, they hit the wall. The platforms have a partial answer: some parametric models with exposed variables. Real engineering under the hood, a more or less friendly interface on top. MakerWorld's Parametric Model Maker (with Fusion support) is a genuine step forward. I've been experimenting with it; the Gridfinity rack I published is an example, and the potential is real.But the limitation is structural. The user can only customise what the creator already anticipated. The form, the logic, the constraints, all decided in advance by someone who knows what they're doing. The user is a passenger, not a driver. And the community of creators who build these models is smaller than it looks. Attrition is brutal. Someone discovers they can model, publishes a few things, gets little visible traction, and quietly disappears. Those of us who've stayed are genuinely exceptional — and I mean that in the literal sense. We are the exception that the system depends on but cannot reliably reproduce. AI generative tools are the obvious next candidate. Text-to-3D, image-to-3D, the forms they produce can be remarkable. For decorative objects, sometimes genuinely surprising.But here's the problem: current AI is fundamentally stochastic. It generates geometry from statistical patterns. It doesn't know what a tolerance is. It doesn't know that the wall thickness matters, or that the gap between two mating surfaces is measured in tenths of a millimetre and determines whether the thing works or binds or falls apart.You cannot derive physical correctness from pattern-matching alone. The two are different in kind. So where does that leave us?Form generation: AI has it. Deterministic engineering logic: CAD has it. Distribution: platforms have it. Manufacturing knowledge: simulation tools have it. Conversational interface: large language models have it.What no one has yet is the architecture that connects them — the system that decides when to be generative and when to be deterministic, when to ask the user a question and when to apply a rule silently.I've published a longer piece exploring this in detail — the full argument, the map of tools, the research gap, and what I think convergence between the major players might look like. I'd genuinely love to hear what this community thinks. You've all lived this from the inside. What am I missing? → [Full article: The Machine That Waits] Jorge Rui Silva — Mechanical Engineer, Industrial Designer

AI-generated 3D models are advancing rapidly, and for tabletop gaming enthusiasts, particularly those creating Dungeons & Dragons miniatures, the changes are striking. I’ve been tracking AI 3D model development as part of a project to explore how these tools can assist in FDM printing and to demonstrate proper AI attribution when sharing models.AI 3D Models Are Getting SharperRecent outputs from MeshyAI show how much AI has improved. For example, when generating dragon miniatures from a photo prompt, the latest models exhibit better coherence with the original image and significantly sharper details. Previously, AI-generated models often revealed themselves through inconsistencies like extra or missing limbs—but that may soon no longer be obvious if outputs are carefully curated.If you’re trying to avoid AI-generated content, this is a good heads-up: discerning AI models are becoming harder to spot. Meshy 4 vs Meshy 5 vs Meshy 6Why Learning 3D Modeling Still MattersEven as AI tools improve, now is an excellent time to develop hands-on 3D modeling skills. Programs like Nomad Sculpt, Blender, SolidWorks, and Fusion 360 give you capabilities that AI alone cannot match:Create high-quality OBJ files with individually selectable parts for multicolor printing.Fully customize models or enhance AI-generated assets.Explore advanced print techniques beyond what AI outputs are suitable forCreate mechanically precise parts for assemblies or mechanisms  Mastering these programs ensures that, regardless of AI’s progress, you retain creative control and access to advanced printing options. Learning to 3d print along side these modeling softwares can also help teach you design principals to create supportless or more complex models with different properties.AI in CompetitionsWhile AI can be a helpful tool, using it in competitions defeats the purpose of the contest. Competitions are meant to showcase human creativity, problem-solving, and skill. Relying on AI undermines the spirit of fair play and diminishes the value of genuinely handcrafted work.A Note on AI Use and AttributionI regularly use AI to edit and rewrite text, and I don’t oppose AI in creative work. My main concern is transparency: AI should be properly attributed when it contributes significantly to a model. Using AI without disclosure diminishes the recognition of human effort behind a project. Personally, I balance AI assistance with hands-on creation to focus on my passions while acknowledging the trade-offs.The evolution of AI in 3D modeling is exciting, but human creativity and skill remain essential. Whether you’re designing your own D&D miniatures or enhancing AI-generated models, learning 3D modeling today will keep your projects versatile, high-quality, and uniquely yours. Edit: This article was mainly looking at Meshy. For those interested, MakerLab has a similar Image to 3d AI available that is on par with Meshy 5 or 6 generations as shown using the credits from the Points system. Would recommend

How to tell if a model was made with AI  AI models are really starting to infest sites like MakerWorld, Printables, and CrealityCloud. Printables has a way of letting you know what models are AI and what are not by flagging them with “AI”. This is a really helpful and nice tool for those of us who don't like to download AI models, as they can have lots of problems and are generally low-quality. CrealityCloud is festered with these kinds of models, and no way to tell without actually looking at the model in a 3D perspective. Makerworld is also starting to see a huge increase in these models, which are promoted with their image-to-model tools and other AI tools. While I personally don't have any problem with AI models, I do have a problem with people not marking them as AI and/or taking credit for it. So how do we know if a model is AI? Well the first thing is pretty obvious. If it's a model of a person, there's always something that just doesn't look right. Missing or mutilated fingers, contorted features, etc. But how do we really know for certain that a model was AI-generated? There's a pretty easy trick I've learned just by studying models. Take a look at these picture:You can see that the edges of this model are jagged, uneven, and has tons of unnecessary triangles. This is the #1 sign of an AI-generated model. Bambu AI is getting really good these days, so it's getting harder and harder to tell. But this is a bad case of AI generation distortion. By the way, this picture is from my Cucko model if you were wondering. Now take a look at this:This is a propeller piece of my B-17G model here on Makerworld. Look at those edges. All around, they're smooth, defined, and orderly. No extra triangles or cracks. This is a real model.  I hope this helps you if you were wondering how to tell real and AI models apart. Thanks for reading!!

A Deeper Dive: AI Tools and the Future of CreationWelcome to the second part of our exploration into the modern creative process. In Part I, I shared an in-depth look at my personal creative toolkit and the step-by-step journey from a nascent idea to a tangible, 3D-printed object; from that initial meditative sketching phase through CAD modeling to the final printed result.Now, as promised, we turn our attention to a force that is set to redefine many of those very processes: Artificial Intelligence. Beyond its utility in tasks like language translation, AI is stepping into the core of ideation and fabrication. How are these powerful new tools beginning to influence each stage of the workflow I described, and what are the broader implications for the design community and the future of making?AI Meets the Creative Genesis: Rethinking Low-Tech BeginningsReflecting on that initial “low-tech” phase of ideation I described — the quiet sketching with pencil and paper, or on my trusted Boox e-ink tablet — I find myself wondering how generative AI might reshape this deeply personal stage, and yes this device already has some AI integration... That meditative practice I mentioned, requiring a clear mind and intentional isolation from existing solutions, now faces an intriguing paradox.While I currently value that degree of mental isolation to foster unique concepts, the prospect of AI as a brainstorming partner is undeniably compelling. Imagine being able to rapidly visualize variations of rough sketches (but with meaning, since that is not happening at this moment, June 2025), or having an AI suggest unconventional starting points based on a simple prompt describing a functional need. The challenge, for me, would be integrating such a tool without sacrificing the serendipity and organic development that I find crucial in those early moments of creation.Remember those breakthrough moments I mentioned — the ones that happen between 2 and 4 AM or in the shower? There’s something beautifully human about that ‘diffuse thinking’ process I described. The question becomes: can AI complement this natural rhythm without disrupting it? Perhaps AI could serve as a catalyst for those focused work sessions, then step back during the mental downtime where the real magic happens. This is something that keeps me thinking…From Concepts to CAD: AI as Design PartnerMoving into the digital realm I outlined — from refined sketching in Concepts to precision modeling in Autodesk Fusion — the potential for AI integration becomes even more apparent. That seamless transition I described, where “the heavy conceptual lifting is largely complete; now it’s about execution,” I’m sure, will be dramatically enhanced in the near future.At the current pace of technological evolution, I find it increasingly difficult to imagine how large, established companies like Autodesk, will be able to keep up with the rapid developments in AI integration. In my view, what’s more likely is that we’ll witness the emergence of specialized AI-driven CAD companies, sprouting like mushrooms across the landscape, much like what we’ve seen with the proliferation of Large Language Models.Just as the LLM space has exploded with numerous players, each offering increasingly powerful capabilities and appearing from the most unexpected quarters, I anticipate a similar phenomenon in the CAD world, and in fact in all areas of knowledge. These nimble, AI-native companies won’t be burdened by legacy code or traditional approaches, allowing them to build solutions from the ground up with AI at their core.This shift could fundamentally disrupt the established order I described in Part I, where familiarity with tools like Fusion 360 becomes less relevant than adaptability to rapidly evolving AI-enhanced workflows. The question becomes: will the deep integration and ergonomic advantages I praised in my current setup be enough to compete with AI-native solutions that might offer entirely new paradigms for digital creation?AI generated image showing a futuristic scenarioBut for now, let’s imagine AI-driven features within Fusion, that could offer real-time structural analysis as I model, or suggest alternative geometries based on functional requirements. The extensive experience I mentioned with Solidworks, and others; all that intuitive knowledge about creating specific shapes — could be augmented by AI that learns from my modeling patterns and suggests optimizations.Consider how AI could transform my workflow with the 3D Scanner, I rely on. Rather than the current process of scanning, cleaning up meshes, and manually extracting geometry in Fusion, AI could automatically clean noisy scan data, intelligently reverse-engineer complex surfaces, or generate parametric features from scanned meshes. This could make that “glove-like fit” I achieve with printer components even more precise and efficient.Even those premium 3Dconnexion peripherals I praised, like the SpaceMouse Enterprise and CadMouse Pro — could potentially integrate with AI, and with time, learn my modeling preferences and suggest shortcuts or workflows based on the current design context.Manufacturing Intelligence: AI in the Physical RealmThe journey from digital model to physical object that I described — using Bambu lab software suite to interface between virtual and real — represents perhaps the most immediate opportunity for AI integration. That careful consideration of manufacturing constraints I mentioned, thinking about print orientation, supports, and material choices during the design phase, could be revolutionized.Imagine Bambu Studio augmented with AI that doesn’t just optimize G-code for speed and quality, incorporating all known and future techniques, but also predicts potential print failures based on geometry and chosen material. It could suggest design tweaks before the print even begins, analyzing those subtle influences of material choice and model geometry that I noted. This will come too, and sooner than we think. This predictive capability could significantly reduce material waste and hours spent on test prints — addressing that reality I mentioned where “despite all care, everything doesn’t always work as intended on the first try.”The iterative nature of moving from virtual to physical could become far more efficient, with AI learning from each print’s successes and failures to improve future predictions.Content Creation Enhanced: Beyond TranslationIn the content creation phase I described — crafting narratives with rendered images and descriptive texts — AI’s role extends far beyond the Gemini or Claude translation assistance I currently use. That process I outlined of creating renders in Blender while maintaining consistent parameters could be enhanced by AI generating diverse lighting setups or environments automatically.AI could analyze the CAD model’s features to draft initial assembly instructions, or even analyze community feedback on shared projects to suggest areas for improvement or new project directions. This could free up more time for the core design and making activities I’m passionate about, while ensuring high-quality presentation and engagement.The Shifting Creative Landscape: Opportunities and ConcernsHowever, this integration raises profound questions about the nature of creativity itself. That organic process I described for selecting which ideas to pursue — where “the most viable solution emerges somewhat unexpectedly” — could be fundamentally altered by AI’s systematic approach to problem-solving.Take the Tower Crane project I mentioned in Part I. Those initial notes and sketches emerged from a specific personal need or community interaction. If AI had been involved from the beginning, would the solution have been different? More efficient, perhaps, but would it have retained that uniquely human perspective that comes from lived experience?On a practical level, this shift is already visible on 3D model platforms. Decorative objects, being relatively simple to generate with small variations, are increasingly AI-created. However, projects like the functional designs I focus on, or more complex mechanisms, that include the Bambu lab Cyberbrick kit; remain largely beyond AI’s current capabilities. The question is: for how long?The Human-AI Partnership: Amplification vs. ReplacementLooking ahead, I believe creators with the necessary discernment and “mental gymnastics” will use AI as a powerful amplifier, exploring uncharted conceptual territories while maintaining that critical human judgment I emphasized throughout Part I. The key lies in preserving that balance between focused work and mental downtime, between systematic analysis and intuitive breakthrough.The danger I see is the temptation to delegate the entire creative process to machines, losing that essential human element that comes from understanding materials, experiencing failures, and developing that intuitive feel for how virtual designs translate to physical objects.That journey I described from “the initial spark of an idea to the tangible object” relies heavily on accumulated experience, tactile feedback, and yes, those 2 AM epiphanies. The challenge is integrating AI in ways that enhance rather than replace these fundamentally human aspects of creation.Conclusion: An Evolving Creative PartnershipAs we stand at this intersection of traditional making and artificial intelligence, the future of creation promises to be both exciting and uncertain. The tools and methods I shared in Part I form the backbone of my current workflow, but they’re just the beginning of a rapidly evolving story.The question isn’t whether AI will transform the creative process — it already is. The question is how we, as makers and creators, will guide that transformation to amplify human creativity rather than replace it. The path forward lies not in avoiding these tools, but in learning to use them as sophisticated extensions of our creative intentions, much like I described my CAD software as “an extension of your hand.”The future of making will likely be defined not by human versus machine, but by the thoughtful partnership between human insight and artificial capability. And that partnership, like any good design, will require careful iteration, testing, and refinement to get it right. If you like this, check my blog at: https://designrepcom.com/ for more, and updates on my projects…

This article is about starting to design and sharing a step by step tutorial for an easy project you can try to get starting with fast and easy results. Because nothing is as motivating to learn more about designing, as creating a working design in about 30 minutes. This tutorial uses online tools that does not require any installation or payment. The only thing you will need to do is create a (free) account. If you are a minor, ask permission from your guardian befor you do.Design tool used: TinkercadWhat is Tinkercad?Tinkercad is a free, web-based 3D design and modeling tool created by Autodesk, primarily aimed at beginners and educators. It allows users to create 3D models using simple geometric shapes through an intuitive drag-and-drop interface. Why would I use Tinkercad?User friendly, ideally for beginnersWeb-based and free, no installation is requiredIt supports STL, SVG and OBJ files. more on this later. Why wouldn’t I use Tinkercad?Limited in use: when evolving to more complex designs, you will come to know the limits of the softwareLimited in size: the files you import can be max 25 MB and a certain amount of detailNon-organic: organic fluent shapes, will be very limited. What files can I import?STL => basically about every design floating around on MakerworldOBJ => about everything AI generated by Makerlab or other MakertoolsSVG => images converted to SVG What are the possibilities?Well, here is the tutorial part!I wanted to show a simple project that you can try yourself to set the first few steps into designing your own stuff with 100% free tools.Project #1: earringsStep 1: create an image using AIAI used: GeminiPrompt: generate a simple geometric image of the face of an animal, with thick black lines, no background, all the lines are connected with eachother, so you don't have to pull your pen up when drawing the drawingResult: Download the image in .jpg or .png format.Step 2: Convert your image to SVG-fileGo to https://convertio.co/svg-converter/ and chose your image, to convert to svg.Download the newly made .svg imageStep 3: import the file into TinkercadGo to Tinkercad.com, make a free account and open a new projectSelect “import”Choose your .SVG fileand press ImportWait until the image is imported.It will look something like this:Step 4: change the designNow - those would be some huge earrings. Lets make them smaller.Like any other image scaling, you can select the file, and rescale them by standing in one of the corners and simply hold and pull.For this design, which is quite round and bulky, I’m going for about 1 inch x 1 inch or 2,5cm x 2,5 cm for earrings. Of course, if you change the scale, your design just got very very flat.Your design will hang on someone’s ears. It. will. Break.For earrings, I suggest a height of 1,6 mm (0,062992126 inch?).Always use a multitude of the layer height you intend to use. I use a layer height of 0.2, so my height can be divided into 0,2. (That’s 8 layers - see math is usefull)Select the middle cube and change it to 1.6 mm, when pressing enter, the design will change instantly Now, we need something to put our earring hook in.When testing I figured out that an opening of 2 mm is perfect to give it some room to dangle nicely when you walk around. But, we have nowhere to put the hole in. So we add a ring!A ring is basically a cylinder with a hole in.So with our basic shapes, we add a cylinder To make it nice and round, you max out the sides.But, it’s to big. So we make it smaller using the same technique as befor, or we type in the sizes manually.Height: 1,6Width (on both sides): 3,4 mm - we don’t want our ring to break when wearing the earringsNow we need to make the hole.Select your cylinder, copy it (control C) and paste it (Control V). Now you have 2.Change the sides of one to:Width: 2 mmNow, we need to make it a hole.Select the small cylinder, and tab “hole” on the top right corner Next step: putting them together.Select both cylinders (hold in control and click, or click and hold left mouse, and drag the selection over both cylinders)Click the align toolSelect the middle boll on the bottomand the middle boll on the leftSee how they turned grey? That means they are 100% alignedTo merge the objects, click the merge iconClick it, give it a few seconds, and your hole has been made:Now to add the ring to the keychain, you can use the same align tool:You will just have to move the cylinder ring a bit to the top, so you actually have a hole. You can use the arrow keys for this.But notice how your cylinder takes big steps? We want little steps, so we can decide where to put it.Change your snap grid to 0.1 mmNow every step your object takes, is 0,1 mm instead of 1 mm. Select both shapes, and merge/group: Final result:Step 5: Export the design into a STL-fileNow select your design and press “export” and STL Give it a minute, and a freshly made STL-file is in your download map.STep 6: print!Add this into your bambu slicer and check if everything is printed well.Too thin lines will not print.Normally, I would just testprint 1 and change the design. But this will clearly be a success - so duplicate the design so you have 2 earrings. Otherwise you will waste more fillament purging then to actually print. Final resultFinish your print off with an earring hook and tadaa! Your very own earrings.Final notesThis is a tutorial to engage people into using designing software and this method uses (a lot) of short cuts to get a quick, motivating result.This is nowhere near how 99,99% of the designs are made, this is a first small step on a road that is beautiful and fulfilling.This is (obvously) not a design that can be an “exclusive model”, because it is:FlatNot particulary well designedI wish you a lot of fun creating your own designs and seeing everyone grow as a designer. I am loving the journey I am on and wish you all the satisfying feeling that comes when you finally get a design how you want it, or learn a new skill in a designing software ❤️ Looking for a print profile? I have uploaded this design with every setting on free so you can use all the tools available.