That new banner image is AI isn't it?

That can’t be a real photo right? The helmet geometry doesn’t make sense, and the zipper slider just disappears.

1 Like

Looks like it to me. They’ve certainly been playing with generated images in their youtube thumbnails. Or they found a guy who had his hands cut off and reattached on the wrong arms.


For the record: I’m not a fan.


Sheesh those worms are a big pile of trash. Also not a fan. Disregarding the theft and ethical concerns, they just look like wicked unprofessional.

1 Like



  • Maybe, maybe not. Considering what can be done with HDR photos as a base, coupled with endless options in retouching, I certainly can see this as a real picture vs AI generated.

For sure you can get that lighting effect with some combination of modern sensor and editing tech. But there’s something around the eyes that feels off. Could be real, but with the obvious generative thumbnails on some of the youtube videos, I’m in the “this is not a photograph” camp.

I read that shorthand for “scraping a bunch of other people’s copyrighted work off the internet or wherever to use as training data.” The arguments about whether this is the same as stealing physical objects is a major philosophical question right up there with “why is there something rather than nothing?”

Trainerroad’s “all-in” on AI is worrisome at best.

TR is a software, programming, and training company. If they’re not all in on AI/ML they’ll eventually become irrelevant.

I champion TR’s integration of AI/ML, as it has immense potential to improve their product. Additionally, I see no issue with AI generated photos. Not only is it cheaper and faster to get the pick you want/need, it has the ability to be more creative.

What exactly is there to worry about?


The fact that all these algorithms are trained on copyrighted images and using them to create “original” works is arguably theft of other artists work.


Not sure I follow: aren’t music, literature and visual art also chain of influences and flows? Obviously, plagiarism is definite no-no, but is this the case here?

Or from another angle, what about asking AI to update your own original photograph in style of some artist whose work is already in public domain?


AI is fundamentally going to challenge our concept of originality, and it may well challenge even broader concepts of creativity. What worries me is that I suspect it won’t be long before it is all but impossible to distinguish wholly human, part human, and wholly AI generated material (be that images, words, or whatever). It has enormous implications in the world of education, and arguably even bigger ones in the world of employment.

Given all that, TR’s banner doesn’t worry me.

1 Like

It really isn’t.


Yea god forbid a human ever creates something inspired by someone else’s work.

I’m just glad that bike companies never copy each other and that every bike on the market is completely original.

1 Like

It’s the T-Rex arm that bothers me :rofl: (Jonathan)
I reckon the’ve added the jersey and the helmet to an original photo, and as you mentioned the zipper slider doesn’t join up to it’s other side.

In the context of TR, why does anyone fracking care? They are selling training software. Do they really need a photographer and a designer to work all day on such an image when and intern can turn one out in 5 minutes with an AI tool?

Whatever people think about AI and jobs going away and all that, the genie is out of the bottle and, of course, a company like TR is going to use the tools.


There is a fairly obvious difference between being inspired from someone else’s work and using copyrighted material as training data for a neural net.


Is it that obvious though? Are you well versed in machine deep learning?

Stable Diffusion was trained on 2.3 billion images. Any single image or even medium-size portfolio of images was inconsequential.

When you see the model accidentally use the original artist’s signature, or a garbled version of it, or when you ask for an image using their style, it’s obvious.

It’s very much like uploading your training files for years, and someone else downloading them and installing them into their body without any of the work or risks, only the benefits.

I think a little compassion towards these artists is warranted. However, the arts are typically not valued despite their significant contribution towards society.


The reason I care is that if they throw out a crappy generated image instead of sweating the details, either with a photographer or spending some time refining the generative prompt, what else are they missing?

The image pasted in by the OP is probably good enough, but TR has out up some really bad generated images. It strikes me as pretty lazy, and that’s not what I want from this company that’s taking a chunk of my bike money.


So you want them worrying about whether a supposed AI image has a zipper on the jersey vs, working on the actual product you are paying for?