“No site without a visualization any more”

The LINDEMANN Group develops, plans and builds in the Hamburg area and the Elbe-Weser triangle – with its own project development, its own planning and sales departments and its own construction company. For a long time, architectural visualization was the exception there: laborious, expensive, either produced in-house with rendering software or commissioned externally. The company has been working with VisioPlan for almost a year. Tobias Wegener, Head of Planning, talks about the effort it took before, the scepticism at the start, a planning approval process – and where he still sees room for improvement.

The interview was conducted in German by Kevin Fechner, founder and managing director of VisioPlan.io, with Tobias Wegener, Head of Planning at the LINDEMANN Group. This is a translation.

“Black lines on white paper”

Let's start at the beginning: how did you work before VisioPlan came along? Ideally with a concrete example – when you had to visualize a project at an early stage, how did that go, step by step?

In terms of drawings, we basically worked in the classic 2D view – black lines on white paper. For larger buildings we then produced a visualization. We tried out the well-known programs – Lumion, Twinmotion, D5 Render – and then built the visualization by hand to get a first presentable result. That ate into our capacity again and again, especially in the early phase, when we didn't yet own the site or hadn't yet won the contract. And still, you had to go in with very good quality even then.

Did you always handle that in-house?

Both. For projects that were somewhat larger, we outsourced too, absolutely. We worked with different visualizers at all sorts of quality levels. You could tell them apart: who can do photorealistic, who can do high-end? And depending on what kind of building it was, the job went to the right one.

We'd also already spent a good year before that working with other software products. We deliberately specialized and trained staff and invested the time – with D5 Render, with Twinmotion. Because we knew: visualization is going to be a must. You can estimate it: when one of our colleagues sat down to it, she'd spend a week on it before the finished visualization came out the other end and we said: that works.

“A blessing and a curse at the same time”

Let's jump straight to today. Have you fully integrated VisioPlan into your processes by now?

It's completely integrated into our process. That's a blessing and a curse at the same time. There isn't a single site of our own that we plan any more without doing an AI visualization in VisioPlan. And we notice it with external enquiries too: an image to win a contract always beats the client seeing their building only in 2D, in the classic elevation. There's simply much more emotion in it.

And the curse?

That the expectation is there now, internally and externally. When we plan one of our own projects or plan for an external client, it's taken for granted that we show the visualization too. It used to be on top, it wasn't included – and everyone was aware of the effort involved. The projects we deliver in-house we probably couldn't have visualized at this pace at all, and certainly not at this quality. Of course there's a time factor with VisioPlan too, until a visualization is truly finished and matches what the planner has in mind – we all have a little streak of perfectionism. But we make that time, because we see the added value and because the result is right in the end.

“It's a fraction of the time”

If you compare onboarding directly: how long does it take in each case until you've mastered the software and get adequate results?

It's a fraction of the time – both in terms of time spent and the scope of the program, compared with Twinmotion, D5 Render and the like. All the preparation you have to do in other software falls away. If I had to estimate: For a first good result, we certainly need only about 15 percent of the time.

And how long does it take until someone on the team has it down as routine?

You need a good four or five projects that you take all the way through once. Honestly, though, that's hard to put a number on, because we went through the development of the software with you. You get a quick result without much time spent. It only becomes perfect with practice: especially with the authorities or larger buildings, the house has to stand there with what's actually to its left and right – not somewhere out on the prairie. Because people hear about AI everywhere, they now also expect the building to stand exactly where it belongs. On top of that, the data available varies by region – Google Earth covers Hamburg much better than our rural area around Stade.

“The results spoke for themselves”

Was there a moment when you were sceptical? And one when that cleared up?

Not so much scepticism in the narrow sense, because we'd been talking about it and trying things out again and again for a year beforehand. We'd also had contact with other AI software. We got good and less good results with it.

It was more about my department, about the users. At the start, they quite quickly get into a spiral: “This doesn't work the way I imagine it. My building is supposed to look like this and that.” You have to talk to people and make clear that we can't expect the one hundred percent final image, because our own base data simply doesn't support that. That was a hurdle you had to get over once: just see whether we can't set off perfectly well with 80, 90 percent.

There was a certain scepticism, along the lines of: up to now we've always developed visualizations exactly to our plans. The house was built precisely to lengths, heights, window heights and distances – that was the base data the visualization software was built on. And here you hand some of that over, because it becomes a visual model. That takes trust that the AI won't start hallucinating, making things up, changing distances or making things in the image look different from how they'll actually be built – and that you won't have to explain to the client afterwards why it looks different after all. At the start, that kept us from walking into an official client meeting with the first rendering that came along. But that quickly took care of itself: the results spoke for themselves, and we simply stopped questioning it.

What matters to us is that we label it as an AI visualization – that we're open about it and point out that one thing or another in the image may not be 100 percent accurate. But we stand behind 80, 90 percent of what you see in the image.

“It can't be required – but it's very welcome”

You also take the visualizations into the planning approval process. What does that do to the conversations at the building authority and in the municipal council?

We can produce images like these faster and include them with even the smallest enquiry. An authority can't demand something like that, it's goodwill, to help the imagination along a little. Someone who works at the building authority can picture it relatively well. A municipal council, on the other hand, is made up of people from all sorts of professions who have nothing to do with construction in their working lives. For them it's naturally easier to picture a project from a colour image than from a black-and-white plan. So it can't be required – but it's very welcome. The municipality now also asks to see it visually.

And the classic documents – how does the visualization relate to them?

Those are technical drawings. Behind them sits someone who has to clarify building law questions – they don't need anything playful, they need a technical plan to orient themselves by. The visualization comes on top, so you can explain it to them well.

And with your clients?

The feedback there is consistently positive too. Clients know from others that they get a black-and-white drawing with the concept sketch and have to imagine the rest. We manage to build emotion for what they've set their heart on relatively quickly. A familiar picture is there very fast – and you also notice quickly whether we've hit what they wanted or whether we're building castles in the air. With many, the first thing we do is simply put the image up on the screen to create emotion straight away. And the image usually stays up until the middle of the conversation.

I take from this that you have a competitive advantage over comparable companies. Or how would you describe it?

When we look at how the others are doing things – yes, we do have a certain advantage.

“Whoever doesn't use AI becomes inefficient”

Suppose you're talking to a colleague from another company and they ask you: is it worth it?

Not worth it, don't you dare. (laughs)

Suppose you're not competitors – they work in a completely different field?

Then: of course it's worth it. You have to adapt your way of working to the model a little, that needs to be said clearly. You have to rethink once: how do you approach things, and how much courage do you have to invest the time so that a promising result comes out of it.

For example, we go in with just the building massing, because we're at such an early stage that that's enough for us for now. Once we're further along, we go deeper into the modelling too. Then you feel the time spent up front, but it pays for itself through the visualization.

And what I always say about AI in general: whoever doesn't use AI becomes inefficient. That's just how it is. Because with a few tricks you get a good result. You have to feel your way and decide: how much time do I invest, how perfect should the end result be? Sure, it can take longer sometimes. But in this case you can rely on the AI to a fair extent.

“What we're still missing”

Where do you currently still see potential? What would make your work even easier?

It would be great if we could get our own materials in. We have our manufacturers, we have our subcontractors who install certain products – that's where we could fill in a bit more between 90 and 100 percent.

And we want to be able to draw even more on our existing data: on our own materials and standards, which we maintain internally anyway.

Would it help if you could create favourites per project rather than in one general folder?

That would definitely help. You always work within a project. In the end I talk to my planners about the various projects, take a look – and I don't want to have to dig the favourite images out of a jumble first.

“That backs up the visualization in politics too”

Take us through a concrete project: how do you go about it today to get good results?

We have a site in Buxtehude that we'd looked at. In terms of building law, there are certain things to consider: I'm allowed to build on the area up to a certain maximum size and height.

We look at every site individually and have to keep an eye on economic viability. We looked at what's around it and made a site plan from that. That's actually how we always start: this is our classic site plan for the first pass, the first draft – two buildings here, parking there. That already gives the first rough ideas about height and number of storeys – and with that a direction: how big will the buildings be, and how expensive can the site be in the end?

Quick question in between: where does the base map come from?

In Lower Saxony it's free. Through open data we can simply download it from the land registry office. We add the green spaces ourselves, just normally in Revit. That's not much effort, we do it as standard.

Then we look at the lines of sight and start playing around: what could the building look like? There were a few basic ideas and different variants, for the roof for example. And with that we went to the town and said: we have a site here, would you go along with this development? Because we deviate – we're one storey higher, we seal more ground, we have more flats. Can you picture that?

The answer was: in principle, yes. That was the first step done. Then you go through several conversations with the municipality and, in this case, develop a solution together. In urban planning terms the tenor was: we'd like to see this development there, including the front building. We just had to provide justification for why it's permissible to build more there than originally planned. And that's exactly where the visualization backs it up again, in politics, in the administration. We had to make the rear building somewhat smaller; the front one could stay as it was, because people liked it so much.

For us it's about economic viability, but also about sustainability: densification, for more residents and new flats – also to reach the housing targets that all of Germany has. That only works with more flats in less space. And through the visualization and the idea of what the building should look like, we managed to respect the surroundings and the neighbourhood and still create denser housing than originally intended.

You can see it in the neighbouring building: a bit of a pitched-roof theme, a bit of old building stock. And with us, a somewhat more unusual roof form, plus a massing that projects to the front – combining the modern with the classic. That probably wouldn't have come across like that in a line drawing or a run-of-the-mill visualization, especially with the lawns, the cycle areas and the parking spaces. And that then struck a chord with the people at the authorities. There are people in every authority, after all, and they like something – or they don't. That's what led to the conversations continuing at all and to us now having reached a final solution.

“The next step goes inside”

What's next?

So far we've put a very strong focus on the exterior, and we're now getting sensible results there. The next step is interior visualization, especially in sales, when we have properties that are harder to sell. We just had a property where the client wanted a partition wall between the living room and the kitchen. We tried to show him what's possible with two variants: a full-glass wall with sliding elements or a masonry one with a door. It's about catching the last client who still needs help making a decision. It's about emotion, about solving problems.

For that, though, we also need to develop a standard so the images always look the same. That's what we try to do with the visualizations too: that they always have the same mood. We don't do anything in autumn, anything in spring, but actually always the same, and we always try to achieve a consistent, coherent image.

I also find the outlook on automatically generating exposés, microsites and sales documents at VisioPlan exciting: if you prepare a presentation during planning and can pull out two or three finished, designed slides, that makes sense.

And functionally it would help us if you could send a rendering directly as a link to people outside, so they can just take a look – without logging in, without having to dig out the image first. For a while we had the IFC file, or rather the link to it, sent to us and showed the model in 3D to clients and authorities. In high-resolution quality, exactly that would be a hit: take the link, pass it on, the recipient opens it and sees the building from outside – and maybe in future can even walk inside or switch floors off.

Everyone is blown away by these visualizations at first. For everyone it's real added value, because they can picture it instantly.