Jump to content

AI Visualizations - Kitchens


Recommended Posts

  • Vectorworks, Inc Employee

These images are always a mix of wow, and less-than-desired solutions. But this prompt is very consistent.

"Abstract modern interior design of kitchen and dining area in the style, high-end minimalistic architecture, wireframe renderings, soft edges and blurred details, marble floor tiles , white walls with wooden shelves and cupboards, large TV on wall, bar table at one end of long island bench with chairs, detailed architectural drawing"

 

kitchen 1.png

  • Like 3
  • Love 1
Link to comment

Is there any benefit to using AI visualization within VW than some other option?  If I understand correctly the visualization is not based on the actual model geometry and therefore not tied to core VW functionality.  Is that correct?

Link to comment
  • Vectorworks, Inc Employee

My way of approaching these AI images:
- Learning how to construct the prompt
- Creating an image from text only until I understand the results

- Open a 3d mode, use that geometry as base image, and then I apply the same prompt

- Edit, edit, edit, test, test, test.

- Until I'm content with a few I like. 
- Then if the project requires it, I'll spend time with my model and add textures and lights until I get close to match the AI results.

Here is a work in progress from a home in Maryland.

 

AI and vwx file.png

Link to comment

So which came first, the model or the ai render?  Is the purpose of the exercise to design a kitchen or to recreate ai renders?  I'm asking legitimately, not trying to make a snarky comment.  It's because I am struggling to see how ai integrates into workflows.  I would imagine that, as a kitchen designer, I would do my space planning first, with some research on the side to establish mood/materiality.  Groundplan begats elevated model, add materials and details.  So at what point am I handing this over to ai....in the groundplan phase?  The elevated model?  Materiality? 

  • Like 3
Link to comment
  • Vectorworks, Inc Employee

Currently our AI will not design a kitchen. Too many moving pieces. The layout came from my own sketches. Moving and relocating cabinets and testing sizes. 
 

I see AI visualizer as being a part of two phases. When I was just imagining styles and arrangements, that’s what an image by text only provides. Sort of like having a dedicated Pinterest handy. But in this case my prompt is a unique list. Style, colors, items like tv screens, shelves with plants, etc. pure inspiration. 

 

Then comes the 2d and 3d work. The real items in place. Cabinets, fridge, stove, microwave, faucet, island, countertops, etc

 

Once the geometry is in place, I can use AI Visualization but now based on the vectoworks view. At that point is all about testing finishes, colors, props,  lighting, but don’t expect to be a precise image, not a substitution for a rendering. It is just a quick way to produce flash illustrations. I like using watercolors or oil ink at this stage. 
 

Then, when everything is decided and approved, time to use Renderworks. lots of precision there, real materials and lighting location. Great control over every aspect. 
 

And finally lets get those drawings out and all reporting needed. 
 

My objective with these AI posts is sharing prompts and looking for others to share their work and prompts composition. That’s how we all can learn. 
 

  • Like 1
Link to comment
  • Vectorworks, Inc Employee

I thought this diagram would help to explain a bit more my new approach to kitchen project designs.
AI Visualization can help on phase 2, brainstorming style ideas and for getting early sketches out of a simple 3d model.

Design evolution.png

  • Like 2
Link to comment

Although I love this stuff, I am sort of in agreement with some of the others.  Where does this fit in?  

 

I can see how this could bridge the gap between people who can visualize in their heads and those who struggle with that.  That is not an untrue statement, some people simply don't visualize well in their minds eye, but are still very capable designers using traditional methods.

 

 

Link to comment
14 hours ago, Luis M Ruiz said:

Open a 3d mode, use that geometry as base image, and then I apply the same prompt

Thanks for the explanation.  If I understand correctly, in the long run you're expecting your pre-developed AI prompts to replace saved SLVP styles you've created for rendering.  While I don't plan on doing this because I have very limited rendering needs, it seems like a sensible workflow if standard rendering techniques don't meet your needs.  

 

Most of the discussion on the forum is still focused on fantasy images, not on explaining how saved prompts can be used in a professional workflow.  It would be great if you could post finished results ideally with a comparison image of your previously used rendering techniques. 

  • Like 1
Link to comment
  • Vectorworks, Inc Employee
12 minutes ago, E|FA said:

"expecting pre-developed AI prompts to replace saved SLVP styles you've created for rendering."

No, no, no. Like I said, AI images are just extremely flexible but they are not supposed to be a replacement for an accurate rendering. If it works for the client great but if you pay close attention, there are tons of geometry errors.
Picture them as a cool coloring book done in seconds. This is why I am using watercolor sketches, ink and oil, pastel colors, as a way to be forgiven for the loose hand illustration.
I am posting another model. All done in Vectorworks, a final Renderworks and a AI images (sketch style).

kitchen model.png

Kitchen-L-Shape with Island.png

AI kitchen sketch.png

  • Like 3
Link to comment
2 hours ago, Luis M Ruiz said:

This is why I am using watercolor sketches, ink and oil, pastel colors, as a way to be forgiven for the loose hand illustration.

Got it.  I could see a use case of presenting loose schematic design phase images to clients even though I have things modeled to a higher level of detail.  This sets a better level of expectation for the clients in terms of the amount of work already completed.  Crisp renderings of a model can make it seem like a project is much more resolved than it really is.

  • Like 1
Link to comment

@Luis M Ruiz I dig the marker drawing, that's my bread and butter.

 

It's just funny going to all that effort of precision modeling and lighting along with staging the scene with so many placed props only to have AI muddy it up.

You are done at that level of detail, it's just a matter of finishes typically.

 

A nice marker drawing like that should only take 10-15 minutes to render using real or digital markers (ie Copic or Procreate) if you use the model for trace over.

Where is the AI saving time and effort I wonder.  Really, how much time do you have to sink into the model to get these AI results?  In my experiments, it seems like it always takes at least 30 minutes to get something grooving in the right direction.  And then I go back to rendering by hand because it's more fun AND faster.

Link to comment
55 minutes ago, Jeff Prince said:

A nice marker drawing like that should only take 10-15 minutes to render using real or digital markers (ie Copic or Procreate) if you use the model for trace over.

Where is the AI saving time and effort I wonder.

 

I guess it is useful if you don't have any hand-drawing skills. And maybe those skills will become ever rarer as AI becomes more widespread?

 

I like to imagine that somehow, having these skills ("talent", even) might keep just a little bit of an edge over what AI can do but that might be fantasy.

 

The same applies to the design skills themselves (rather than drawing/presentation skills). AI will start making inroads at the lower end of the market first, design-wise - those who churn out variations of semi-standardised designs will probably be the first victims.

Link to comment
24 minutes ago, line-weight said:

I guess it is useful if you don't have any hand-drawing skills.

 

I would hate to be the person who presented a sketchy marker drawing to create the illusion of having artistic skills and not be able to produce the same in person during a client charrette.

  • Like 1
Link to comment
20 hours ago, Jeff Prince said:

 

I would hate to be the person who presented a sketchy marker drawing to create the illusion of having artistic skills and not be able to produce the same in person during a client charrette.

I hear you there @Jeff Prince - however  recently in a zoom mtg client asked for a major scenic addition and after the smoke cleared from the ask, the scenic designer built the piece in 3D in moments in the zoom. That was pretty cool. Definitely still not as fast nor as attractive as a good hand sketch.

  • Like 1
Link to comment

I'm just unsure that this is where ai is needed.  The developments in real time rendering and openGL have come so far that it's not hard to imagine how in the very near future we can make changes to our models and see the results faster than you can type in the ai prompt, even with artistic flairs.  I need/want ai to do more background/workflow like issues of the design phase.  

 

 

  • Like 3
Link to comment

Unless someone if working/playing with AI, it will never get better. Luis is doing us all a favor by working to see what is possible.

 

And where I see AI helping in rendering is in added entrourage. Do you really want to have to put books, plants, statues, clocks, etc. into your VW Model in order to get a fully populated scene?  I certainly don't.

 

The visualizer is doing what it can today. If all of the people who say "that is not what I want" were listened to and work stopped on the visualizer, we would never get improvements.

 

If it does not click for your brain and workflow then don't use it. I seriously doubt that the inclusion of the Visualizer is impacting the development of what you do want.

 

My $0.02

  • Like 4
Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...