This AI agent is designed to help generate exploded views of devices and objects in a more streamlined way.
The process starts by bringing in an image of a product. The agent then prompts and generates a new image showing the object in an exploded view, revealing internal components and structure in a clean, readable layout.
One useful aspect is that it doesn’t strictly require reference images of the interior. The agent can infer a plausible internal structure on its own, but you can also provide reference images or schematics if accuracy is important or if you want more control over specific parts.
Beyond still images, the agent also supports animation. You can animate the transition from the intact object to the exploded view, reverse the sequence, or create intermediate shots like zoom-ins, angle changes, or partial disassembles. These clips can be stitched together within the agent and exported as a single video.
This makes it well suited for product breakdowns, technical explainers, marketing visuals, and educational content.
This is the agent we use, and below I’ll walk through how it works and how to use it.
Deconstructed Product Agent:
https://lnkd.in/eX4QAnvY
Have you seen this affect before? Let me show you how it's made. These are a few examples of what we've created using our Destructive Product Agent. It works great for gadgets, devices, and mechanical objects, but you can also experiment and push it further to see what else is possible. I'm going to walk you through how this agent works from start to finish and show you how to create this exploded effect quickly and easily without a complicated setup. Let's jump right into it. All right, so here we are on our agent, and all we really need is an image of whatever it is we want exploded. So I'm going to go ahead and use an image of a Nintendo Switch 2. And I'm going to drag that into the chat right there. And I don't even have to type anything. You don't have to put no prompts. You can literally just click down here on this arrow. Or just press enter. Alright, so it analyzed the image and it's telling you everything it sees. And then we're going to go ahead and click on exploded image, and now it's going to generate the image for us. So we got some results here. And if you're happy with them, you can just roll with this. But there is a little trick you can do if you want the pieces inside to be a little bit more accurate to what's really inside of a Nintendo Switch or whatever device you are using. You can search online and try to find images where a device has been. Broken apart or sometimes people will break apart a device and then lay all the pieces out and then take a photo of that. So in this case, I do have that. So let me show you what I found. So I found this online. As you can see, there's a bunch of pieces here and we can feed it this image. We could have done this from the beginning, but I just wanted to demonstrate the difference between giving it one image of the device versus 2 images of the device and the insides of the device. So I'm going to go ahead and bring this image into the chat right here and I'm going to let the agent. Know that I also wanted to look at this image as a reference image to know what's inside the device. So now I'm instructing the agent use this image as a reference of what's inside the Nintendo Switch. I'm going to press enter and alright, and this is what it gave us. And as you can see, there are a few extra details that weren't there before like this part of the controller right here. You still you see that it's. Right here, and some of these other things that are inside are a little bit more accurately represented. There's the image of the battery right there and you can see the batteries there too. So there's a little bit more extra things in there that weren't accurate and the original image that we did, there are a few things that it that it did get right, but not everything. So if you want a little bit more accuracy, it might not be perfect, like it won't be like exactly every single detail, but it's worth trying to make sure that you get a lot of those internal parts as accurate as possible if that's what you're going for. And also if you're not happy with this, you can always retry and see what else is going to give you. So yeah, there's different ways you can do this. So like you can do different iterations and this is the great thing. You can instruct the agent, hey, I want something completely different, a different angle, different perspective. I want to correct something. So once you're happy with one, we want to let the agent know which one we selected. So I'm going to click on here is where you see this little quotation mark and it's going to bring the link down here and I'm just going to tell it, this is the approved image. Let's animate it. Once you're ready to animate, you're going to have three options, three different models that you can choose from. You have failure version 2 you. Cling 2.5, you have C dance Pro 1.5 and they all do things differently. So what? So what Kailua you get more precise controlled movement, but the limitation is that it's lower overall quality compared to clean right? So you got cling, higher visual quality and more creative motion can be slightly less precise and introduce creative variation. That's the limitation. Then you got C Dance Pro 1.5, which generates sound, which could be great because when you have all these. Mechanical parts moving around and you don't have to do the sound design for that. I mean, that's gonna save you a lot of time. And it it does, and it produces more animated mechanical motion. And I'm going to show you examples of these right now. Limitation movement is more stylized and less physically restrained. So earlier I generated 3 videos using the three models just so you can see the difference, right? This first one, let's try to guess what model it's using, right? Right. It's mechanical, it has sound. So it's gonna be C dance pro, which worked really well with this. I I really like this one here. I'm gonna just guess just by that by the way it animates. OK, it's really smooth. And a hand comes in at the last minute. I'm gonna guess this one as cling and. This one actually is not clean. This is likely. This is likely haul. I would guess this one is clearing. The quality is way better. It's cleaner looking. So yeah, so that's kind of the differences between these. So you have these options. I think it just depends on what is it that you're working with. If it's very mechanical, I do like the way C dance animates and I love the fact that it has sound, but sometimes clean gives you some really good results too. So I would I would kind of play around with. Those two mostly, and if you're still not getting good results, I would try kind of as last resort, I would try haul O. Here we have our generation. I had to iterate it because my 1st result didn't come out the way I wanted it, but this is going to give you an idea of how C dance works if you're going to notice it right away. It moves really cool, right, like it's very mechanical. I like I like the way it moves. I love the sound effects. The sound effects are great right there. The the colors changed right, So I had to tell the agent, hey, can you make sure that the colors do not change, don't change them to orange and like blue right here. So I think it just got confused because I think some of the Nintendo switches have these colors and it just probably did it by default. So it might be wise to instruct it. To keep it like. The original image right? It adds the sound effects for you here. After I did the the duration it corrected it for me and it's going to animate slightly different. Right. And yeah, so it did that. And right now I'm also generating with Cling 2.5 and Halo just to see the difference. OK, we have Kailua right here. Yeah, very, it's pretty smooth. It's not the perfect, I would do a few iterations to get it just right, but it's it does OK and then we have clean right here. Let's check this out. OK, it's doing a rotation. It's not very accurate, but it's very smooth nonetheless. So you have to instruct the agent don't do a 180 spin or asking not to spin and just to keep it in place as it. It breaks apart. My favorite one so far has been Sea Dance. But this might not always work for you. You might want something a little bit smoother. So these are the options that are available to you. So my recommendation will be that you get the end frame, the exploded frame, just right. Get it to a place where you're happy 1st, and then you can start generating it with video. Because if you're not happy with the final frame, then the animation is just not going to work for you. So make sure that you have the final frame looking just like you want it. And everything else will fall into place, right? And then you can also keep going. You can have it like let's say this one, for example. Let's say you really like this one and just say, you know, click on here. And this quotation mark, it's gonna bring the link down here. So I'm going to tell the agent, can you extract the last frame of this? And let's have another image generator where there is a very tight close up from a different angle. So I generated this for me, and I'm going to go ahead and click on this little quotation and then this one. Then just ask it animate these using start and end frame. Let's go with C dance. Alright, so let's go ahead and test that. Oh. OK, that was a cool transition. However I do not like how it keeps turning these into orange and blue controllers. This could be a sea dance issue because I actually want to test it out. Try this with Kling and also with Halo. I believe this is specific thing with two dance, but we'll test it right now with cling and haul. So that's something to keep in mind. That's something that you have to start to kind of learn from each model. They all have different ways of doing things and the way it. Interprets and understands things. So with Sea Dance I know that I have to be a little bit more specific with things because then it could be. For example, since they already changed the controller colors before, I should always let it know not to change the colors every time I generate. So that's something to keep in mind with. This model. Alright, so let's check out Cling and haul. Alright, so let's check this one. This is Kailua. Nice not to bad. OK, and this is cling. All right, this one feels a little bit better just because the way it comes into you know this frame of the where the screen is supposed to be at just seems like it just a lot smoother here. Just it feels like it kind of just warps into it, which is not as nice. But that's the thing about playing right again, these all have their strengths and weaknesses and you just kind of have to. Figure out which one do you want to go with and which one you're happy with. So this is the thing about models that that they are imperfect. There's only so much we can do. To make the best results. So a lot has to do with the models themselves and you guiding them into the right direction. So this is how you use this. Since I really like this one, I'm going to go ahead and say select that one and say I want this to be my second video. Can you get my first video and this video and stitch them together? My first video is the one I generated with C dance. Alright, so now let's check this out. Let's see what how this turned out, alright? Hoover Mechanical and then it goes into this. Very cool, yeah. And you can even say, hey, I want that to go a lot faster. So the only problem with this is that if you are using C dance, it has audio and if you speed it up, it's going to speed up the audio and it's going to sound weird. So just keeping that, keep that in mind. If you do want to speed it up, you can do that. If you didn't want to have it transition into something different, for example, like you just wanted to have it open up once like this, right? You can either just reverse it. Normally, which is it's just going to reverse the clip or you can generate it in reverse. So in this case, I'm just going to ask it to reverse it. Can you reverse this video? I should have asked it. Can you reverse it and stitches so that it just does it all in one shot, right? So here it is. It's playing forward and then once it gets fit in, it's going to go back and reverse. And this time it doesn't have the sound. So that's a bit unfortunate. So that's one way you can do it. So you can get that clip and just reverse it and stitch it together and it goes right back to the first frame. But if you really want the reverse effect to be a bit different of how it starts and not just like a reverse of the original video, then you can always do it generated reverse, which essentially just switches the first and last frame and generates that. The benefit of reversing the original clip is that it's. 3 And when you have to do it the other way, then you have to generate it using a video generating model. Just to show you what I mean by this, we have our reverse clip, right? All it is is just reversing the original clip, which is fine, but it could be kind of like playing and people can tell it was just reversed. But then you have this kind of reverse where it's doing like a little bit more unique movements going back into place, which is. Very cool too, right? You can do the normal reverse or you can do the generative reverse. So you have that choice. And that's how you use our Destructive product agent to create this exploded effects. Go ahead and try it out for yourself and see what you can make. You've got a lot more tips and tricks coming your way, so stay tuned. Until next time, take care.
I’ve been experimenting with Runway, an AI for motion design, over the last few days and it’s been interesting to see what’s possible.
My key learning so far is that the text-to-video doesn’t work well without an image reference. To get good results, you need to prompt it like an art director, setting the scene and mood. Once you do, it’s surprisingly strong at adding camera movement, textures, lighting and environmental details.
Also AI isn’t always worth it for animating product UI, I can often do that faster in After Effects. But it’s exciting to see how it can turn images into video and create more realistic scenes for performance ads or social media.
How to Create Consistent 3D AI Animations in Under 20 Minutes
Stop letting character inconsistency ruin your AI videos.
Many creators believe a single prompt is enough to create a perfect animation, but this often leads to messy, inconsistent results. The real secret to professional-looking 3D content isn't years of technical training, it is having the right AI workflow.
Here is a step-by-step breakdown of how to create a consistent 3D animation story from scratch:
1. Plan and Script (ChatGPT) Don't make it up as you go. Use ChatGPT to generate a story structure and write detailed image generation prompts for every scene. This ensures you have a clear roadmap before generating a single pixel.
2. Solve Character Consistency (OpenArt) The biggest struggle in AI video is keeping characters looking the same across different shots.
• Use OpenArt’s Character 2.0 feature with the Nano Banana Pro model,.
• Start with a detailed description (e.g., a robot named "Bolt" in 3D Pixar style) to "train" the AI on your character's appearance from all angles.
• This ensures your character retains the same design and proportions in every generated scene.
3. Animate Static Images (Kling 2.6) Turn your consistent images into video using the Kling 2.6 model (available inside OpenArt).
• Pro Tip: When prompting for video, describe the motion (e.g., "camera slowly pushes in," "smooth natural animation"), not just the image content.
• Set clips to 5 seconds for the best stability.
4. Add Audio & Edit (ElevenLabs & CapCut) Generate professional voice overs and sound effects using ElevenLabs to drive the emotional narrative,. Finally, layer your clips and audio in a free editor like CapCut for a polished finish,.
You don't need complicated 3D software to tell a compelling story. With this workflow, you can produce a complete animation in under 20 minutes.
#AIAnimation#GenerativeAI#OpenArt#3DDesign#CreativeWorkflow#DigitalStorytelling#TechInnovation#ContentCreation
For years, as a character designer, my work was mainly supported by static imagery: renders and illustrations capturing a specific moment in time. It’s still a powerful way to communicate, but with the arrival of artificial intelligence, something shifted in how I present my projects.
Today, AI has allowed me to give my characters movement and a sense of action, adding a new layer of life and narrative. Being able to show how a character moves, reacts, or interacts creates a more immediate connection and brings dynamism to a project, even at early stages of the creative process.
AI-assisted animation hasn’t come to replace anything, but rather to reinforce the way my creations are presented. It has become a valuable tool for exploring ideas, prototyping, and communicating more clearly, helping viewers get closer to the true intention behind a character.
At the same time, this does not diminish the value of professional 3D animators in any way. There is a clear difference between animations generated as part of experimentation or training and the deep, intentional work of a skilled professional. Acting, timing, emotion, and storytelling remain essential human skills and continue to be needed in the market.
AI can help move a character, but it is still human talent that gives it meaning, intention, and soul. Understood this way, technology doesn’t compete with creativity, it amplifies it.
Here are a few examples of the dynamism and playfulness that have allowed me to bring my 3D characters to life with fun, expressive actions ✨✨✨
#AI#3D#characterdesign
The motion design industry is dying.
Seedance is just a hint of what's coming.
But this is why I don't think we're completely f*cked.
I dont want to hear your excuses about but AI cant do X or Y, If you havent learnt by now that its continually improving at an ever increasing rate I dont know what to tell you.
The end goal is clear: a real-world simulator.
In that simulation you can create characters assign them personalities, move cameras, swap out and manipulate 3d objects, hell you could even rewrite the laws of physics - control literally everything.
Now imagine you had those storytelling superpowers. What would you create with them?
The scary thing for creatives is that everyone will have those tools and they won't be technical or complex to use. I can't see us dutifully laying down keyframes in the future.
Just because everyone can create doesn't mean they want to. There will always be creators and consumers.
And let me tell you this doesnt happen in isolation to the creative industry, AI agents will replace a multitude of jobs. At some point we’ll be in post-labour economy where your cognitive or physical labour will be worthless.
So what becomes important?
Well, it's the same thing as today - attention.
Attention = power and status.
The ability to direct and control the masses.
Motion design is the art of storytelling, advertising, education and persuasion.
Focus on these and we’ll survive the next evolution of the tools.
Ignore AI at your own peril.
Stop designing images. Start animating attention.
I’ve always loved the posters from Verdens Skove — bold, simple, powerful.
So I took the one hanging on my wall…
and gave it a pulse.
Using ComfyUI + Wan 2.2, I turned a static poster into a subtle moving animation. No After Effects. No complex timeline. Just AI as a creative amplifier.
If you’re working with visuals, marketing, campaigns, events, storytelling, this changes the game.
And when you combine it with image tracking in Hololink ApS, suddenly every static image can become interactive and alive in the real world.
A poster isn’t just a poster anymore.
It’s an entry point.
#AI#GenerativeAI#Wan22#ComfyUI#CreativeAI#MotionDesign#DigitalArt#VisualStorytelling#AIGenerated#CreativeTech#MarketingInnovation#FutureOfDesign#verdensskove#hololink
Is AI replacing motion designers?
Not really… it’s expanding who gets to be one.
Tools like Vibe Motion let you create real motion graphics from a single prompt no heavy timelines, no months of learning curves.
That animation idea you ignored because it felt “too hard”?
Now it’s doable in minutes.
AI isn’t killing motion design.
It’s unlocking it for more creators.
Anyone else experimenting yet? 👀
#motiondesign#graphicdesign#AI#generativeAI#design#futureofdesign
Bring Characters to Life with Grok AI 🎥✨
Guys, animating characters has always been a pain. Frame by frame, tweaking every movement, making it look natural… it takes forever. Weeks for just a few seconds of smooth animation!
Here’s the game-changer: Grok AI. This thing actually understands motion, anatomy, and even emotion. You can feed it a scene, a script, or even just audio, and it will make characters move naturally — like they really exist in that world.
Why it’s amazing:
Speed: Turn ideas into animated sequences in minutes, not weeks.
Consistency: Every gesture, every posture looks real, scene after scene.
Creativity: Experiment with crazy movements or styles without animating every frame.
Flexibility: Works for 2D, 3D, or mixed animation projects.
Imagine: a brand mascot reacting to a voiceover, an educational character gesturing perfectly, or even an NFT character moving dynamically — all without killing yourself in post-production.
This isn’t just AI, it’s Grok AI understanding your story and your characters — making them come alive effortlessly.
💬 How would you use AI-generated character animation in your projects?
I’ll be honest. I was terrified what an emerging technology like #AI can do it our industry (Animation). As we all know AI has the capacity to generate 1000+ images in a minute but can it really translate your brand’s idea in 5secs?
Mountain Flower Animation Pvt. Ltd. We have accepted AI as our creative partnership which allows us to combine an emerging technology to the heart and soul of animation. This has made our work faster, better and qualitative.
Here are 3 strong differentiators of Mountain Flower Animation positions AI+Animation
👉 AI for Speed. Artists for Soul.
We use AI to accelerate motion, backgrounds, and transitions — while our animators handcraft expressions, emotions, and cultural nuances that make your brand story truly human.
👉 AI Visualisation. Artist-Led Storytelling.
AI helps us generate rapid visual concepts in minutes. Our creative team then transforms them into cinematic storyboards aligned with your brand’s voice and vision.
👉 Automation Meets Signature Craftsmanship.
AI handles repetitive frames and lighting — our artists focus on key poses, hero shots, and final polish to deliver studio-quality storytelling at startup speed.
So, here we are your innovative craftsman to convert your brand stories to life. Contact us for best visual storytelling.
Rajat DivateRavi Bevinagidad#animation#visuals#storytelling#craftsmanship
🎬 𝗔𝗻 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁 𝗶𝗻 𝗔𝗜-𝗣𝗼𝘄𝗲𝗿𝗲𝗱 𝗔𝗻𝗶𝗺𝗮𝘁𝗶𝗼𝗻
I’ve been spending time exploring animation with the support of AI 🤖 trying to understand not just the creative process, but also how technology can meaningfully speed it up.
One key learning from this experiment was realizing just how much time AI can save ⏱️. Tasks that earlier was taking 2–3 𝗱𝗮𝘆𝘀, such as finding characters, converting them into 3D, creating voiceovers, sourcing backgrounds, and animating everything. can now be completed within 1–2 𝗵𝗼𝘂𝗿𝘀 with the right AI tools 🚀. This shift alone has been a real eye-opener for me.
This short video is one such experiment 🎥. It’s not perfect, and that’s okay. More than the final output, the real value for me was the learning that came through the process. Working on this helped me grasp the fundamentals of animation, while also showing how AI can make ideation, creation, and iteration faster and more accessible 🚀.
𝘍𝘰𝘳 𝘵𝘩𝘪𝘴 𝘱𝘳𝘰𝘫𝘦𝘤𝘵, 𝘐 𝘶𝘴𝘦𝘥 𝘢 𝘤𝘰𝘮𝘣𝘪𝘯𝘢𝘵𝘪𝘰𝘯 𝘰𝘧 𝘵𝘰𝘰𝘭𝘴, 𝘦𝘢𝘤𝘩 𝘱𝘭𝘢𝘺𝘪𝘯𝘨 𝘢 𝘴𝘱𝘦𝘤𝘪𝘧𝘪𝘤 𝘳𝘰𝘭𝘦 🛠️:
• 𝗔𝗱𝗼𝗯𝗲 𝗘𝘅𝗽𝗿𝗲𝘀𝘀 – for quick visual design and video editing 🎨
• 𝗖𝗮𝗻𝘃𝗮 – for layouts, graphics, and ready-to-use creative assets 🧩
• 𝗢𝗽𝗲𝗻𝗦𝗵𝗼𝘁 – for video sequencing and basic animation edits 🎞️
• 𝗘𝗹𝗲𝘃𝗲𝗻𝗟𝗮𝗯𝘀 – for AI-generated voiceovers 🎙️
• 𝗚𝗲𝗺𝗶𝗻𝗶 – for ideation, scripting, and creative support 💡
#AI#Animation#CreativeAI#LearningJourney#Experimentation#AdobeExpress#Canva#ElevenLabs#Gemini#OpenShot
I spent the weekend comparing HeyGen to three other AI video tools that bring photos to life. 🎥
**If you're choosing between HeyGen, Runway, Veo 3, or D-ID in 2025… here's what actually matters.**
🎯 **Quick Take:**
- **HeyGen:** Best for talking-head videos & explainers. Upload a photo, add a script, and get multilingual lip-synced avatars in 175+ languages. Super easy interface.
- **Runway Gen-4:** Best for cinematic motion. Animates your photo with artistic control for short clips (~5–10 s), visually stunning but no native audio.
- **Google Veo 3:** Best for photo stories with sound. Animates photos into ~8 s clips and generates ambient audio or speech, integrated in Gemini.
- **D-ID:** Best for budget-friendly avatars. Fast social clips with talking photos, lower cost entry. Practical rather than cinematic.
🔍 **What I Noticed:**
- **Audio Integration:** HeyGen and Veo 3 handle audio natively; Runway requires a separate tool.
- **Ease of Use:** HeyGen and D-ID are drag-and-drop simple; Runway offers more creative control but has a learning curve.
- **Video Length:** Caps at 8–10 seconds per clip; longer content needs stitching.
- **Price:** D-ID is the budget-friendly choice. Runway and Veo 3 prefer subscription with credits.
📌 **Recommendation by Goal:**
✔️ **Multilingual explainer videos:** HeyGen
✔️ **Cinematic photo animation:** Runway Gen-4
✔️ **Photo + audio storytelling:** Google Veo 3
✔️ **Quick social avatars on a budget:** D-ID
If you're crafting marketing videos or training content from photos, HeyGen's voice library and template system make it unbeatable. If you want your photos to move like a movie scene, Runway is your go-to.
**Which one are you leaning toward? Or are you already using one of these?**
#AIVideo#HeyGen#RunwayML#VideoCreation#ContentCreation#AITools
When I try to use to sign in this says