Westlake Publishing Forums

The depot is mine, everything else is AI. Microsoft Copilot copied the depot and Google Gemini created the rest. It took a few days, a lot of false starts, and corrections and cleanup in Affinity Photo to produce this image. AI pictures never turn out as you anticipate.

Another example is the street scene. I wanted my Swayze Hotel model (second from left) to be the center of attention. Gemini had other ideas. At least the image required only two attempts.

You never know what AI will do and that's why I currently consider it a toy.

Russ

It's been a little slow around here so here are some fairly recent time wasters I came up with using Google Gemini. All models are HO. So far it seems the best way to approach Gemini is to accept anything reasonably credible and modify what you dislike in Affinity Photo or Photoshop. Oh, well, it's better than nothing. -- Russ

Here are final four. All HO except the On30 Climax. You never get what you expect with Gemini. Sometimes it faces your locomotive in the wrong direction or changes the perspective and usually muddles the lettering and numbers. At least it frees you from having to create scenery. -- Russ

Russ, feel free to give me a lashing and send me to the corner, but I really hate those AI images.

This goes far beyond adding a background. There is literally nothing left that connects the images to scale models. They say nothing of the quality of the scale models that served as visual prompts for the AI.

I absolutely believe that the starting point were models, but I have no way of knowing where the models ends and the AI begins. And the models have definetively been "improved" by the AI.

These illustrations have nothing to do with photograpy. The images are strinking at first look, but they all have that "Uncanny Valley" feel that almost all AI-illustations  suffer from.

This is getting depressingly close to the AI-generated "modelling" that is suffocating almost all FaceBook modelling groups.

Im beeing harsh, Russ, I know. But I feel very strongly about this. I fear a world where it is impossible to communicate great modelling using photography because everybody have gotten used to AI-perfect illustrations with hardly any connection to the physical models that inspired them. Any mediocre modeller can use AI to make  Ready to Run models look like masterpieces.

On a positive note, I think the disussion is very interesting, and this thread is great as a "lab experiment" in a controlled setting. But this virus is roaming free on almost every other corner of the Internet.

And Russ, I of course I have sinned, too. A while back I used AI with a prototype image as starting point for "making" this diorama.



Its instant gratification, its kitchy and looks sort of nice,  but at the end of the day it has absolutely no value to me compared to my actual modelling.

Quote from: Hauk on June 29, 2026, 10:58:56 PMIts instant gratification, its kitchy and looks sort of nice,  but at the end of the day it has absolutely no value to me compared to my actual modelling.

Agreed (philosophically anyway)! It's an insidious thing. Just because you can doesn't mean you should.

I use AI quite a lot.  It is extremely useful for debugging computer issues, writing code, and supplying rounded answers to questions that I am casually interested in. These are all things that save me time and where the results either do not matter that much OR their performance can be objectively verified.

Things like figures are slightly different. There's a task that I could do manually, but I am not especially skilled at it, it would take a long time and, realistically, I would never get to it. Possibly I would buy in figures from another supplier that would not fit the need as well. So AI is meeting a need I'm not practically able to meet in another way, for a result that is subjectively better. It can be argued that AI is just a tool, but tools need to be directed while AI, by definition, does not.

AI is somewhat like the food additives that are particularly prevalent in the US.  They are ubiquitous, stimulate, are habit forming and do not contribute to well-being (but do improve corporate bottom lines). Their presence distorts our view of reality and our expectations of how things should look.

Certainly a manipulated background (and foreground) can make a model look more convincing and AI will deliver an attractive image. Does this add anything to our appreciation of the model? I think not. Yes, the new picture has immediate impact, but the role of the actual modelling in that is diminished.

As modellers we are always trying to simulate reality in miniature. AI readily improves the subjective simulation, but adds nothing to the modeller's efforts.

Very handy though...



 

I agree with both of you. Completely. The photos do not look real; they look like a background from a golden age Disney cartoon. Please recall I prefaced my post by calling the images "time wasters" and they certainly have nothing to do with the reason this forum exists: Model building.

While I deliberately avoided asking the question, I was hoping somebody would comment analytically on what seems to be a slight improvement in what AI (in this case Google Gemini + Nano Banana) is churning out now instead of condemning a technology even AI itself fears.

AI's developers are building massive facilities across the U.S., hogging crucial water resources, electricity, real estate, and RAM in their mindless rampage of greed. Overall we may be better off without the technology but soon people will embrace it just as they embraced computers so they could "surf the net". Today "consumer" AI is mainly a free or low cost toy to create demand; soon we'll pay for it whether or not we want it. Corporations already are replacing humans with it. Try calling Apple tech support if you have any doubt.

The point of my post was to stimulate activity on the forum while, at the same time, to entertain you.

Oh, well.

Russ

A background project That was posted on the forum several years ago seems relevant again if only for the sake of comparing a much earlier technique to AI options.




 

This was among the first projects I did for the HO scale NEB&W. It was a two part project. Part one started with a Laser Arts kit for the Shoreham Covered Bridge in Vermont that I was given to build for a new branch line being added to the layout. The kit depicted the bridge after restoration in the early 1980s for use as a hike & bike trail. I backdated it to the 1954 time period of the layout.

Because the club wanted the scene to be a photo focal point, Part two meant accurately recreating the 1950s background scene behind the bridge, long lost to time, underbrush and trees.

The club archives had lots of photos of the scene from the time period. This view, taken by John Gardener around that time served as the template for creating the background view.


This is a test shot of the bridge in place with some temporary foreground greenery and the printed background taped temporarily very closely behind it.


Here is another archive photo of the scene used as reference.


Creating the backdrop was my first attempt at using GIMP (like an open source PhotoShop). it was a loooong learning process. I used a Google Earth map view and landmarks in the photos to locate Gardner's camera position. and how much of the Google map view woulwas captured in the Gardner photo.

Then I erased everything from the Gardner photo except the distant treeline. That treeline was duplictaed as closely as possible by combining bits from many digital photos I took.
.

The open field in the foreground and some of the far right end of the backdrop were created both with bits of other photos I took and cloning of parts of them. The final backdrop was 72 inches long by 14 inches tall.

The resolution was 100dpi, which was a big image file for my computer at the time, but it was fine for the distant scene. Unfortunately the entire NEB&W was dismantled before the scene could be completed. The color test shot abve is the only photo of the almost completed project.

What took me months to do ("learning" as I worked) could now perhaps be sort of rendered by AI in a few minutes at most, but the bridge and the background scene were great projects to do. (Side note: Amazingly I managed to salvage the bridge model, part of the framework it was installed on and the printed background scene).

Quote from: finescalerr on June 30, 2026, 01:24:27 PMThe point of my post was to stimulate activity on the forum while, at the same time, to entertain you.

I think you managed that quite well!

Just to be clear, my frustration is more directed towards myself than you and your friend. I really struggle to find the balance between
using AI in a creative, meaningful way and just wasting my time and natural resources. For my professional work as an architect, I have totally capitulated. I use AI all the time. Im expected that I do so, and it is unavoidable in a business that is getting more cut-throat every day.

But with modelling, it is another story. To me, modelling is much more than a hobby. It is a field were I have some talent and a lot of dedication to be at a level that attracts some positive attention. I think I would have been a modeller even if nobody saw my work, but getting som recognition for what I achive DO  mean a lot. Especially the "peer reviews" I get on a forum like this.

But why would we admire work that took almost no effort to create? I think that our enthusiasm for the AI generated figures seen in another thread is that we knw how much effort and talent that would have been neccesary to create the figures with traditional means. When this type of magic becomes routine, I fear that the fascination will be gone.

What I admire is talented modellers putting in a lot of effort even if they use advanced techniques like AI. When the end result is outstanding despite the hi-tech tools. The balance is crucial. Or rather, the scales must be tipped so that it is obvious that most praise is due to the modellers efforts and not the technology.

I fear the day when nothing stands out because we all use AI. Pardon the amateaur philosophy, but I sincerely feel this would amount to an existential loss.

Quote from: Hauk on June 30, 2026, 11:05:18 PMI fear the day when nothing stands out because we all use AI. Pardon the amateaur philosophy, but I sincerely feel this would amount to an existential loss.

True.

The kind of hobbyist who would abuse AI already exists. He is in a majority of perhaps 100,000 ... and he is uninterested in our forum. It is pointless for us to worry about his effect on "the hobby" because that hobby is fundamentally different from ours.

People who choose to create things themselves, to scratchbuild rather than buy or have a machine create something for them, are always around and they'll always be a small minority. Our forum consists of only a few dozen people with only a handful actively building something at any given time. None of us abuses AI.

What does that tell you?

Russ