N Gauge Forum

@Ensign Elliott pointed me in the right direction with this, although I used Copilot AI rather than ChatGPT to generate a snow scene from a picture of my Moorpark layout. Here's the original and the AI versions:

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-031225171309.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156528)


(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-031225171425.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156529)

It's not bad at all. The AI image has few anomalies but you wouldn't know if you didn't see the images next to each other. It couldn't read the "Moorpark Junction" sign clearly, so it generated one of its own. It cropped the left side of the picture, maybe because it was out of focus and it didn't recognize the double-track mainline and made one single track instead.

Took about two minutes to generate the new image. Pretty amazing, really.

This is the problem with some of these AI enhancement tools, they don't just work on your image they selectively replace stuff.   

I noticed this when I gave one of the cheap online photo enhancers a blurry old photo of two of my long missed cats, hoping it would simply do a bit of sharpening. What I got back was a cute photo but clearly were no longer my cats. 
Worse still, the so called "try it for 10p" offer had actually set up an unauthorised recurring monthly payment for £49.99 from my PayPal account, which fortunately I spotted and cancelled after PayPal prompted me to review it!  (I then found several people reported this thievery on TrustPilot).

@ntpntpntp Send me the picture of the moggies if you like and I'll run it through Photoshop - I've got access to the full-blown Adobe Creative Suite here.

@LASteve   actually now that you mention it, so do I have access to Adobe Creative Suite.   I currently have in my possession a recently deceased modelling friend's Macs and other devices while we deal with his estate.  He was a school reprographics tech.

This A.I stuff is scary.. its so easy to generate images that look so realistic its hard to tell if they are real or not.  Granted you may find errors, but given its "learning" ability it won't be long before those errors get corrected.  Problem is people do believe what they see... an AI generated video of the late queen dancing in what looked like a care home environment still convinced lots of people that she was still alive !

I've not used it for image work, but I have for programming ESP32 devices.  I described what I wanted the project to do, and what sensors to use and it built the program in a matter of minutes.  I then spent around 8 hours getting it to add functions and cosmetically look a little better - some attempts it broke the code, but it then went back and fixed it.  After a days work I had a fully functional project that would have taken me 3 - 6 months and with a human mentor to debug my code.

The image above looks amazing, very life like and something that could be used for a personalised Christmas card.

Using Gemini.Google I took an image posted by @Bob G  and 'winterised' it.
Note the original has a stain top left and it's now gone :)

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-041225123904-156537577.jpeg)
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-041225123904-1565371058.jpeg)

In the November 2025 edition of Railway Modeller, Ian Blenkinsop describes how he took a black-and-white photo, colourised it using computer software, and then scaled it to OO gauge and printed it to use on a building for his layout Cambois Bay.

I am interested in trying to colourise some old B&W photos and wondered if NGF members had any recommendations for a suitable software. 

I would also be interested in software that can remove "cracks" and lines on old B&W photos.  My interest is primarily on colourising railway photos but also possibly some old family photos. 

There seems to be many software companies offering colourisation, such as HitPaw, NCH, liftapp, Canva and many, many more.

I would be interested to hear of any real-world experiences. Also, because I am somewhat thrifty, I would be looking to start with free software (not a free trial), if possible. 

Any advice will be much appreciated. 

Ian

@icairns - I took a B&W photo of a warehouse, loaded it to (free) gemini.google and gave the prompt "colorize this picture". Orginal and resulting AI below:


(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225165611.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156560)



(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225165641.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156561)

Took less than a minute to product the render. Pretty impressive, eh?

Quote from: LASteve on December 04, 2025, 04:57:40 PMTook less than a minute to product the render. Pretty impressive, eh?

Yes, very impressive!  I am going to give it a try. 

Any other other ideas still welcome.

Ian

And one more ...

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225170509.webp) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156562)


(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225170552.png) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156563)

Quote from: LASteve on December 04, 2025, 05:06:29 PMAnd one more ...

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225170509.webp) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156562)


(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225170552.png) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156563)

Did you tell AI which colours to apply?

If notit's done a pretty good job in my opinion.

Quote from: port perran on December 04, 2025, 05:12:27 PMDid you tell AI which colours to apply?

If notit's done a pretty good job in my opinion.

Nope, a three word prompt "Colorize this image". I thought the color rendition was pretty amazing both with the loco (the red buffer beam especially), the carriages and the distinction between the brick station building and the corrugated concrete outhouse.
Looking closer at the tender, it looks like it added the British Railways emblem.

Quote from: Newportnobby on December 04, 2025, 12:42:37 PMUsing Gemini.Google I took an image posted by @Bob G  and 'winterised' it.
Note the original has a stain top left and it's now gone :)

As has the curve of the track with the vans in the distance.. which is a lovely example of why it's cute for christmas cards and shouldn't ever be used to "enhance" stuff for research or reference.

Quote from: port perran on December 04, 2025, 05:12:27 PMDid you tell AI which colours to apply?

If notit's done a pretty good job in my opinion.

For most surfaces an image a little context and some guesses on the shading is enough to accurately colourise it. My brother is totally colourblind (as in he sees only grey shades although he assigns "colours" to them by learning for compatibility with normal humans). He can accurately recolour most of the black and white things we've asked him about just fine. This has been doable fairly effectively before large language models got used to do it 1000 times less efficiently 8)

My brother was a godsend watching snooker on the black and white tv.

Quote from: EtchedPixels on December 04, 2025, 05:38:16 PMMy brother was a godsend watching snooker on the black and white tv.

"For those of you watching in black and white, the pink is next to the green." - Ted Lowe, BBC

John P

Quote from: jpendle on December 04, 2025, 06:24:56 PM"For those of you watching in black and white, the pink is next to the green." - Ted Lowe, BBC

"For those of you watching in black and white, Spurs are in the all-yellow strip." — John Motson. I'm pretty sure I was watching MOTD when he came out with that. Spurs were playing Leeds away, so everyone except the 'keepers were wearing all-white on our telly.

@LASteve had a great result in reply #9 but it failed me using......

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-041225123904-156537577.jpeg)

with the instruction to colourise it........

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-041225184659-1565641960.png)

I did another "Winter" scene using Gemini. This one preserved the original more accurately, but it doesn't look quite as atmospheric.


(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-041225191922.png) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156567)

Quote from: jpendle on December 04, 2025, 06:24:56 PM

Quote from: EtchedPixels on December 04, 2025, 05:38:16 PMMy brother was a godsend watching snooker on the black and white tv.

"For those of you watching in black and white, the pink is next to the green." - Ted Lowe, BBC

John P

I did indeed get to see that moment live on TV

@LASteve

Wow, thanks, Steve.  Your suggestion of using Google Gemini has me fascinated! 
Here are two examples of colourisation.



I like the very subtle colouring including the red buffer beam.  I have no idea what colour the prototype loco carried.  The second attempt was a little more problematical.



I took the above photo of 70039 Sir Christopher Wren waiting to reverse onto Carlisle Kingmoor shed on 12 August 1965.

You will note that for some reason the steam from around the front cylinders has been replaced by an incorrect smokebox design.  Now why would it do that? 

I colourised another photo of me standing in front of LNER Class J27 number 65819 at South Blyth shed on 17 July 1965.  I was wearing a grey windcheater, sweater, and blue jeans.  The loco turned out alright but I was transformed into wearing a brown suit! 

Ian

Looks like something funny went on with the driver, too

Since everyone else is at it I thought I'd better have a go so I fed Copilot this and asked for a snowy scene;

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225073853-1565652396.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156595)

and this is what it came up with

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225073853-1565652295.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156594)

It looks quite pretty until you look more carefully - not only has it downgraded the crane's capacity (and lost the builder's name) but, to add insult to injury, it's turned the steam locos in the background into a blasted weasel   ???

Yeah, I'm over it. Not so smart, eh.

But it has managed to re-rail the loco for you!  :no:

 :laughabovepost:

I never noticed that! :smiley-laughing:

I noticed in Ian's photo, that the Britannia is one of the few with the rare 2-6-2 wheel arrangement.  Quite unique!

Quote from: PLD on December 05, 2025, 08:26:39 AMBut it has managed to re-rail the loco for you!  :no:

But has made the driving wheels different sizes :D

Quote from: Newportnobby on December 05, 2025, 12:05:39 PM

Quote from: PLD on December 05, 2025, 08:26:39 AMBut it has managed to re-rail the loco for you!  :no:

But has made the driving wheels different sizes :D

Yes, it seems to be undecided as to whether it's a 2-6-0 or a 4-4-0 with an overly long coupling rod.

At least it hasn't put people in the tender  :smiley-laughing:

Quote from: PLD on December 05, 2025, 08:26:39 AMBut it has managed to re-rail the loco for you!  :no:

And get rid of the troublesome points.

Here is another attempt at colorizing my Britannia photo.  I changed the command in Google Gemini to be more specific as follows: 

   "Colorize this photo. Do not change any white steam - leave as original."

This is the result.


This manipulation looks much closer to the original B&W photo (see post #19, above) except AI has now added a fireman! 

Ian

Tried a few more when the daily limit on Copilot had cleared.

This wasn't a bad try, although it did derail the Jinty. It really fouled up the signal in front of the platform end so I erased that;
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192056-1566161889.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156617)

Another pretty good one, although it didn't seem to know about sheeted wagons - the one in front of the logs is more resembling a small tender now.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192055-1566142259.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156616)

Now this one isn't too bad at all, it's had a good try at the nameboard.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192055-1566141641.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156615)

Can't see much to complain about this one.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192054-1566121871.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156613)

It had trouble with the nameboard here but at least it didn't replace it completely. It also struggled with spoked wheels but it didn't try and change the sizes.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192054-1566121682.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156614)

Another very good try, my only real criticism here is the tunnel mouth.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192053-156611118.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156612)

It had a reasonable try with this one, but the tearoom roof was obviously asking a bit much.
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_7182-051225192053-1565952460.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156611)

Might be 42 degrees in Wollongong at the moment, but not in Bealman's backyard! :D

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/255-051225215454-1566192070.jpeg)

We seem to have sprung a leak in the clubroom roof and Heatherley has copped it

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-051225222031-1566202015.png)

Anyone remember this photo from the gathering of forum members at TINGS in 2016?

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_262-061225144958-1566291379.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156630)

Far too much time on my hands. I think I need to get on with my layout.

Jeepers! :goggleeyes:
I'm glad I was taking the photo!

Quote from: Newportnobby on December 06, 2025, 03:36:50 PMJeepers! :goggleeyes:
I'm glad I was taking the photo!

I'm sure AI can rectify that for me. Unfortunately, or fortunately for you, I have used up my free quota for today  :*(

 :laughabovepost:  :smiley-laughing:

Now that IS an impressive use of AI. ;D

Are you able to put names to Tutu's?

Wondered what Mid Albion may look like so went to Co-Pilot, loaded the pic and here we have before:

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/1058-061225223747.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156643)

and after:

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/1058-061225224100.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156644)

think I may go playing with a few pics of the home layout

Not to be outdone by @Trainfish here's York 2023. Notice that Mrs B sensibly refused to be part of such nonsense.

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/255-061225225615-156637312.png)

Oh gawd, what have I started?

Well Mick got his tutu! ;D

Phew, that's a lucky break, I had just left the group to go and listen to a Ben Ando talk!

Tim

As a bit of a luddite on AI I thought I'd have a very quick dabble and added snow to four pictures of James Street. I'm impressed at the speed of it but I would say about 50% impressed with the results. One appears to have have very little snow, one has been hit by a blizzard and the other two are really good. Probably down to the quality of the images used?

Here's the best two....

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_3680-071225101015-1566538.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156654)

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_3680-071225101014-1566482002.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156651)

Less successful....

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_3680-071225101014-1566481095.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156652)

And a real blizzard.....I don't think anyone is going to get off that train any time soon!

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_3680-071225101015-1566521608.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156653)

Dave

Quote from: Southerngooner on December 07, 2025, 10:15:38 AMAs a bit of a luddite on AI I thought I'd have a very quick dabble and added snow to four pictures of James Street. I'm impressed at the speed of it but I would say about 50% impressed with the results. One appears to have have very little snow, one has been hit by a blizzard and the other two are really good. Probably down to the quality of the images used?

The results do seem to very variable, I often got different results from the same image and instructions.

I have a suspicion that it doesn't just work using the image supplied but also by looking at other, possibly similar. images that it's seen elsewhere - that's the only reason that I can see for it replacing steam locos with diesels or coaches, both of which I've had it do.

Certainly the Google Gemini is fallible. I tried to add snow to a pic of my Blue Pullman and it removed the driving car and replaced it with an odd BP middle coach!

Quote from: Moonglum on December 07, 2025, 09:07:35 AMPhew, that's a lucky break, I had just left the group to go and listen to a Ben Ando talk!

Tim

@Moonglum Tim, you got away very luckily in that I have the same pic with you in it. Trouble is my AI programme doesn't understand "Add pink tutu to the 4th person from the left". It appears it just cannot count and puts it on the wrong person :doh:

This is what happens when you go to a beer festival in Devizes !

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/230-071225151144.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156657)

That was SO last year. At least we all made the effort this year. Not sure about those wellies though:

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/medium_262-071225164925-156654521.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view;id=156658)

Thanks John @Trainfish,

I've been waiting for that one all Weekend. I was then trying to think of something funny but I don't really have the words although my wife's first comment was that my legs look better than they do in real life  :doh:

Cheers, weave  :beers:

and they used to say that the camera never lies :(

Quote from: Browning 9mm on December 08, 2025, 08:58:26 AMand they used to say that the camera never lies :(

And that's the main issue.  All these images above have been in good faith, but to the eye they all look plausible. AI can be used to generate deep fakes and for of an older generation who are so used to believing that a photo can't lie often won't accept it.

It was the same when visual effect / CGI software became in reach of the average person and would run on a home PC (albeit a top end one) and people started producing UFO images. There was one that showed two craft fly over a tropical island, and my late father who was stationed at the same island after the war swore blind it was real.

People have used AI to generate topless, or even pornographic images of well known Hollywood stars - Its been almost impossible to undo the damage it done to their reputation as the images have to be examined through meta data to prove they were fake.  But by then the damage is done. 

Quote from: Browning 9mm on December 08, 2025, 08:58:26 AMand they used to say that the camera never lies :(

They were lying.

Sir Arthur Conan Doyle was taken in by (non AI) fake photos of fairies supposedly taken in 1917.
https://en.wikipedia.org/wiki/Cottingley_Fairies

Quote from: Jim Easterbrook on December 08, 2025, 11:42:15 AM

Quote from: Browning 9mm on December 08, 2025, 08:58:26 AMand they used to say that the camera never lies :(

They were lying.
Sir Arthur Conan Doyle was taken in by (non AI) fake photos of fairies supposedly taken in 1917.
https://en.wikipedia.org/wiki/Cottingley_Fairies

Not really the same. The camera didn't lie, it took genuine photos of fake subjects - unlike AI manipulated or, pre-AI, Photoshopped alterations after a photo's been taken.
 

Quote from: Jim Easterbrook on December 08, 2025, 11:42:15 AM

Quote from: Browning 9mm on December 08, 2025, 08:58:26 AMand they used to say that the camera never lies :(

They were lying.

Sir Arthur Conan Doyle was taken in by (non AI) fake photos of fairies supposedly taken in 1917.
https://en.wikipedia.org/wiki/Cottingley_Fairies

Nooooooo....that was real  :D

Real estate agents have been producing photos of properties for sale that look bigger than they actually are for ages. I know because they did it to my townhouse, and that was back twelve years or so.

I do think that Malc-c has a very valid point, though.

Quote from: Malc-c on December 08, 2025, 11:31:38 AMAnd that's the main issue.  All these images above have been in good faith, but to the eye they all look plausible. AI can be used to generate deep fakes and for of an older generation who are so used to believing that a photo can't lie often won't accept it.

People have used AI to generate topless, or even pornographic images of well known Hollywood stars - Its been almost impossible to undo the damage it done to their reputation as the images have to be examined through meta data to prove they were fake.  But by then the damage is done.

Couldn't agree more.  I hate to use a slippery slope argument, but that's what this is.  A normalisation of a fundamentally disruptive and damaging technology.  While I get that photo-manipulation has always been a thing, the scale of what you could do was pretty limited and required specific skills the vast majority of people have never had.  By definition AI images are dishonest and available to everyone, and while we can all see the problems with the attempts at adding snow, that won't be the case for much longer. 

I come here to see and read what other people build, warts and all, indeed the warts are the real benefit as they show me that my limited skills are no different than other peoples, and so I feel part of a broad accepting community of varied skill, practice and approach.  As more people use AI manipulation (indeed as more and more gets 'built in' to the technology we use) this will diminish and expectations will change, standards will converge to a highly manipulated norm and everyone will be worse off for it.

Sorry for the rant, I could go on for days like this!  Think its time at least for us to have a forum rule that any AI manipulated images have to be labelled as such as I know I can't stop the march of AI as much as I might want to

Hmm... Having seen certain of these AI generated/manipulated images that have been included on this thread, I've decided that it's simply all far TUTU much for me... :o

Just meant to be a bit of fun. It didn't snow on my layout but I was quite happy when it AI-did.

I work in an industry when both generative AI and agentic AI is a concern, but we're working to embrace it, and police it with appropriate guardrails. I'm on an AI council to examine and approve or deny AI-driven projects.

"Make Moorpark snowy" as an AI project would not have received a whole lot of attention.

I'm sorry this thread became a little dark, not the intent.

Depends on what you think of blokes in Tutus, I suppose  ;)  ;D

Quote from: LASteve on December 09, 2025, 03:02:03 AMI'm sorry this thread became a little dark, not the intent.

Good comedy is often on the edge of dark thoughts.

The thread has been very thought provoking, as well as being the first time I've seen some "AI" photos without the usual faults like people with 3 arms or 6 fingered hands. (That'll be the "Norfolk'n'Good AI".)

As stated, it's a bit of fun, but it has pushed me to look into it. As a result, I'll be taking a 4F, BG coach and snowplough to the club on Wednesday with a view to creating a view of the club layout which I can add snow to and make an e Christmas card to send out to all our members.

Granted its all been a light hearted bit of fun, but what if rather than tutu's the instruction was to make the group in the photo naturists and remove all clothing...possibly with genitalia blurred as its a family forum... It might make a few people chuckle, but those "victims" in the image might think otherwise..

What's concerning me is how quick AI is progressing now that its freely available.  Soon it will be designing the next level of data centres to take AI to a new level.. some of those sci-fi films like I-Robot could well become reality.

As I said in my posts with the James Street snow pics I'm a Luddite with regard to AI. However, I do think it is possible to go a bit over the top with complete rejection of it. Yes, it COULD make you naked, it COULD be used for harm, but so can a car, a gun, drugs, social media, war, etc, pick your own choice of harmful things that we have become accustomed to. Keep your head down, don't go broadcasting your life on social media and try to avoid getting whipped up into a frenzy by click bait headlines on TV, internet or newspapers.......thats my recipe for a better life!

Dave

As they said in the excellent Full Monty movie, "Just think of the most boring thing you can think of"

A Dire Staits double album!

Quote from: Bealman on December 09, 2025, 11:12:36 AMAs they said in the excellent Full Monty movie, "Just think of the most boring thing you can think of"

A Dire Staits double album!

Heathens  :smiley-laughing:

Quote from: Bealman on December 09, 2025, 11:12:36 AMA Dire Staits double album!

See what happens when you use AI spellcheck? Not to be trusted ...

Ooops, it's worse than that - that was my own doing, completely unassisted by spell check or AI! :-[

In a possibly vain attempt to get this thread back on track, here is a photo of LNER Class J72 number 68736 that I took at Newcastle Central Station back in the winter of 1962 when I would have been nine years old. 

This locomotive acted as station pilot at York before transferring to Newcastle in 1961.  It has been painted in NER green and carried both the NER emblem and the BR crest. 

Dapol is planning to release this model in the same livery and I have already got mine on pre-order. 

Thanks to AI, I have been able to colourise this photo to remind me what the prototype looked like and I will probably make a print of this colour photo for my railway room.  This is what I see as one of the benefits of AI. 

It is a pity that I did not manage to get the entire loco in the photo.  Well, I was only nine! 

Ian

P.S.  I did specify that the loco be colourised in NER green but the result is possibly a little dark.  But I still like it!

Nicely done! Which AI tool did you use? It's interesting to compare the merits/demerits of each.

Quote from: LASteve on December 09, 2025, 10:06:46 PMNicely done! Which AI tool did you use? It's interesting to compare the merits/demerits of each.

@LASteve

I took your advice in Post #7 above and used Google Gemini. 

However, I found that several iterations were necessary to achieve the final result. 

I started off by issuing the command "Colourise this photo."  The result was a loco in LMS red (similar to Newportnobby's DMU photo in post#16). 

So, I became more specific with my next command: "Colorise the locomotive in this photo North Eastern Railway green." 

This was a better result but the boiler remained uncoloured. 

I then stated (using the colourised photo): "Colourise the locomotive boiler North Eastern Railway green but leave the smokebox and chimney black." 

This produced a good result but I was not satisfied with one detail, so I gave the instruction to "Colourise the lion inside the emblem on the right of the number 68736 dull red."  And, Hey Presto, it did it! 

I tried several other commands to colourise the locomotive (a) Saxony green, (b) lighter green, and finally, (c) lime green.  None of these had the desired effect of getting the loco colour a bit lighter.

But overall, I am pleased with the result.  Probably more details could be done with additional specific commands.

Thanks for the idea to use Google Gemini, Steve! 

Ian

It's interesting that it's followed your instructions to the letter - the surrounding areas including the train are virtually untouched.

I'm currently 2/3rds way thru an OU photography course "Creating and sharing better images" - which explores the use of Digital Darkrooms and digital cameras. As an established 35mm 'operator' this has been mind boggling to find how far from reality one can actually 'push' a photograph!
Without using AI, which must have parameters or algorithms limiting the range of tweaks, there are ample opportunities to 'push' photos from their origin into another dimension. I'm using an iPad and the Apple photos controls but it's not limiting the scope to change images.
You might need an Apple Pencil if you want to colour bits where the AI does this for you, however the pencil can be loaded with more precise colours from a large palette. It could boil down to 'horses for courses' as I can appreciate the coloured versions of B&W images and films as an improvement in the presentation.
As the layout progresses I'll try and produce some improved photos to accompany postings - OU course ends in January '26 and layout activity recommences :thumbsup:

Quote from: Greygreaser on December 10, 2025, 08:14:00 PMWithout using AI, which must have parameters or algorithms limiting the range of tweaks ...

It's more the opposite, you can get AI to pretty much replace the original with something entirely different; it's all about the prompts you give it.

It's interesting that you're using Apple Photo for some judicious editing. I've done the same thing in Photoshop to take "noise" out of vacation pictures - for example I had a photo of St. Paul's Cathedral taken very early in the morning as the sun was coming up from across the Millenium Brdge. Unfortunately it looked like two tower cranes were growing out of the dome so I cropped those out. The latest version of Photoshop has what they call "Generative Fill" which does the same thing with AI, just a whole lot quicker.

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/6889-101225214557.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156742)

Look Ma, no cranes!

Quote from: Greygreaser on December 10, 2025, 08:14:00 PMYou might need an Apple Pencil if you want to colour bits where the AI does this for you, however the pencil can be loaded with more precise colours from a large palette.

I was never any good at staying within the lines in 'painting by numbers' :-[  :*(

Quote from: Steven B on December 10, 2025, 09:00:08 AMIt's interesting that it's followed your instructions to the letter - the surrounding areas including the train are virtually untouched.

Yes, I noticed that myself.  I plan to keep experimenting with this photo to see what enhancements can be made.

Ian

Now this is a very good use of AI.  Granted there are AI artefacts such as 5 legged horses if you look closely, but what a wonderful way to bring drawings to life

The chap on One Tree Hill in Greenwich changed his stockings before the close-up. Maybe he got the white ones muddy :)

  "Unfortunately it looked like two tower cranes were growing out of the dome so I cropped those out. The latest version of Photoshop has what they call "Generative Fill" which does the same thing with AI, just a whole lot quicker."
[/quote]
I did the same with a whole batch of photos having discovered too late! a thumb print on the lens :doh: here's a doctored one
(https://www.ngaugeforum.co.uk/SMFN/gallery/156/8822-111225205106.jpeg) (https://www.ngaugeforum.co.uk/SMFN/index.php?action=gallery;sa=view&id=156754)

Quote from: icairns on December 09, 2025, 07:38:01 PMThanks to AI, I have been able to colourise this photo to remind me what the prototype looked like

Although I'd point out it's redesigned parts of the locomotive like the safety valves. Not something you can trust versus using re-colouring tools.

It's a nice little loco though - I have a couple of old white metal kit ones of it. Nice to know Dapol will be making another tender motorising kit with throw away body though  ::)

Quote from: EtchedPixels on December 11, 2025, 09:35:39 PMIt's a nice little loco though - I have a couple of old white metal kit ones of it.

@EtchedPixels

I am not aware that anyone made a white metal kit of an LNER J72.  Do you have any more details?

Ian

I had a pic of my Kawasaki and just gave the instruction "Set the motorcycle in a country winter scene" and got a pretty decent response from Google Gemini. Then opened it with 'Paint' and set a textbox to create a seasonal card for the Club forum. Pretty pleased with it.

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-131225084638-156763556.jpeg)

Quote from: icairns on December 12, 2025, 12:56:21 AM

Quote from: EtchedPixels on December 11, 2025, 09:35:39 PMIt's a nice little loco though - I have a couple of old white metal kit ones of it.

@EtchedPixels

I am not aware that anyone made a white metal kit of an LNER J72.  Do you have any more details?

Ian

Had a dig - the kit was it seems an ABS J63 originally on a minitrix chassis and someone has converted it a bit (safety valves, no cylinders, box at front. It's a very old (and hardly fine scale) kit in either form 8)

So apparently there was no J72 kit. My error.

Quote from: EtchedPixels on December 13, 2025, 02:56:04 PMHad a dig - the kit was it seems an ABS J63 originally on a minitrix chassis and someone has converted it a bit (safety valves, no cylinders, box at front. It's a very old (and hardly fine scale) kit in either form 8)

So apparently there was no J72 kit. My error.

@EtchedPixels

For those interested in the early history of British N gauge, two manufacturers made white metal kits of the somewhat obscure Great Central Class 5A that became LNER Class J63.  The class consisted of only seven locos and all had been withdrawn by 1957. The availability of an outside cylinder chassis may have been one reason these kits were developed.

In 1967, Minitrix produced a kit to fit on their German T-3 0-6-0T chassis.  This was a stop-gap until the RTR Minitrix dock tank became available.  (This was also the first N gauge loco I ever bought and the first white metal kit I ever attempted). 

To allow the motor to fit, the kit has an overscale cab and the general features were more freelance than prototypical in appearance. 

In the 1970s, Beaver made a more prototypical white metal kit.  This was designed to fit the Minitrix dock tank chassis (which had a smaller motor than the German 0-6-0T chassis).  This kit was acquired by Adrian Swain and became part of the ABS range. 

The photo below shows my Minitrix version on the left and the ABS/Beaver version on the right. 

Ian

Quote from: Newportnobby on December 13, 2025, 08:59:43 AMI had a pic of my Kawasaki and just gave the instruction "Set the motorcycle in a country winter scene" and got a pretty decent response from Google Gemini. Then opened it with 'Paint' and set a textbox to create a seasonal card for the Club forum. Pretty pleased with it.

(https://www.ngaugeforum.co.uk/SMFN/gallery/156/264-131225084638-156763556.jpeg)

So that is your "Mobility Scooter" then?   :D  :D

Yeah - I popped over to Lapland but no sign of Sanity Clause :no:

Well it's obvious that it flies - there's no tyre tracks in the snow  ;)

I have hummed and hahhed for days over whether to post this because of the content.

But there are a heap of AI posts spreading false information and images of the tragic Bondi Beach massacre last Sunday.

Definitely the dark side.

Quote from: Bealman on December 19, 2025, 07:05:56 AMI have hummed and hahhed for days over whether to post this because of the content.

But there are a heap of AI posts spreading false information and images of the tragic Bondi Beach massacre last Sunday.

Definitely the dark side.

It's that sort of dark action that then leads to accusations of 'crisis actors', which just stokes further hatred. Really makes me angry.

Hi All,

Hope everyone had a good Christmas :D . Now that we're in the wilderness between Christmas and the New Year I thought I would have a go at "enhancing" some of my photos.

I have a few threads on building kits, which I hope help some decide whether or not the kit is for them. Sometimes in concentrating on painting the kit as best I can, I neglect my hands  :-[

I tried Google Gemini for the first time this afternoon, and here are the before and after clean up photos ;)

(https://www.ngaugeforum.co.uk/SMFN/gallery/157/5755-281225185920-1570041562.jpeg)

(https://www.ngaugeforum.co.uk/SMFN/gallery/157/5755-281225185921-1570051562.jpeg)

And one more ....

(https://www.ngaugeforum.co.uk/SMFN/gallery/157/5755-281225185929-157006937.jpeg)

(https://www.ngaugeforum.co.uk/SMFN/gallery/157/5755-281225185914-1570031733.jpeg)

Off now to take some photos of my washing up to see just how magic it is  :D  :D  :D

Happy New Year  :claphappy:

All best wishes

Kevin

 :beers:

You can a job on a skin care advert, Kevin! ;D

For hands that do buildings and are as soft as your face.............use....... :music:

For a minute there I thought you were gonna say "As soft as your ar...."  :D

Quote from: Bealman on December 28, 2025, 10:25:05 PMFor a minute there I thought you were gonna say "As soft as your ar...."  :D

Nah. It was actually mild, green, Fairy Liquid with Nanette Newman :)

I actually remember that advert!

Lyrics by U2:  "We're mild and green, and squeaky clean..."