Vintage dial check using photoshop.

[interestedwatcher]

The TLDR, last four photos show the outcome of using photoshop to check dial authenticity.

Forgive me if this has already been discussed, but if people are interested I'm happy to make a youtube video of how to do this, it takes about five minutes.

Lets say you've spotted a dial online that you quite like, it's very rare (and therefore expensive). You don't know the seller, and you want to do your due dilligence.

For this test I selected a Gilt underline Rolex 1675, to both have the Gilt and the underline, it should be between 1961 - 1963 from my understanding.

So I've found a dial, and I quite like it, time to do some digging for a good comparison dial. I've found that HQ Milton is a really good reference place for this, sure enough I locate a 196x Gilt underline dial, and a good face on shot.

I open both the image of the dial in question, and the reference image. I crop the dial in question to a smaller size, and I copy it as a layer onto the reference dial - this is the slightly tricky part, I free transform the dial to line up the text / coronot / outside track etc onto the dial as best I can (it is important to have the image on the same plane, as any slight deviation can obviously skew your results slightly)

I then drop the opacity of the top layer down to 40 percent, and adjust the size, and match it as best I can onto the reference dial, then you adjust the opacity from 0 - 100 percent in 10 percent increments, and look at the results.
0 percent



10 percent



20 percent



30 percent



40 percent



50 percent



60 percent



70 percent



80 percent



90 percent



100 percent



So I quite like this dial. It looks good

To highlight what this might look like for example if you compared a Gilt underline dial to a later dial, this is how it would come out using this method. I have lined up the 'GMT Master' text on the top line as best I could.

0 percent



10 percent



20 percent



30 percent



40 percent



50 percent



60 percent



70 percent



80 percent



90 percent



100 percent



So the juxtaposition here of the two dials (one correct, one not) really highlights it I think

Reference dial



Subject dial overlayed 80 percent



incorrect Reference dial



subject dial overlayed 80 percent



I hope this has been helpful in someway.

[BigBlue1]

Very cool. Interesting. Thanks for sharing. No way am I doing all that, too much work. Computers are not my skill set.

[interestedwatcher]

This example is a Mk3 DRSD dial transposed over a Mk1



Now a Mk3 transposed over a Mk3 (on both I lined up the 'SEA' and let the rest lie where they fall.



Both images are at 66 percent opacity.

[Jackie Daytona]

That’s a pretty cool and innovative idea for authenticity checks. Very nice. Well done

[linesiders]

Yep - been doing this for years ; )

[swish77]

I love stuff like this, so well done. But how is it better than just using your eyes to pore over dials? What is the advantage compared to old-fashioned studying and comparing? Maybe I'm missing something?

[interestedwatcher]

Originally Posted by swish77

I love stuff like this, so well done. But how is it better than just using your eyes to pore over dials? What is the advantage compared to old-fashioned studying and comparing? Maybe I'm missing something?

I guess it's just Another tool in the armory. A very useful tool though, especially in seeing those slight differences that may be hard to see.

[swaini3]

... and save your eye sight lol

[zapokee]

Very nice idea.

Knowing how much variation there was with Rolex/Tudor consistency back in the 60s and 70s, I wonder how concerning it should be if overlaid images don't match up exactly.

[interestedwatcher]

Originally Posted by zapokee

Very nice idea.

Knowing how much variation there was with Rolex/Tudor consistency back in the 60s and 70s, I wonder how concerning it should be if overlaid images don't match up exactly.

I think some slight variation is okay, mostly because the photos may be on slightly different angles etc, but it's important to find the exact dial you're looking at to compare - that isn't always easy with Singer, Beyler etc all making dials around that time.

[rootbeer7]

Clever, but wiidth of each word of text, relative position on the dial and font serifs are probably better indication particularly on older dials made by many manufacturers. This idea is great for later mass-produced models definitely.

[interestedwatcher]

Originally Posted by rootbeer7

Clever, but wiidth of each word of text, relative position on the dial and font serifs are probably better indication particularly on older dials made by many manufacturers. This idea is great for later mass-produced models definitely.

Agreed - this was the final check I was doing after establishing the dial types, fonts and italicisation etc.

[zapokee]

Originally Posted by interestedwatcher

Agreed - this was the final check I was doing after establishing the dial types, fonts and italicisation etc.

Very good work, and fascinating potential.

Based on this kind of comparison, how hard would it be to have an app that you can upload a watch photo to and it would tell you whether the piece is bona-fide or questionable? I imagine it would only require a relatively minor database of known genuine dial/hand/case images for reference.

[linesiders]

Originally Posted by swish77

I love stuff like this, so well done. But how is it better than just using your eyes to pore over dials? What is the advantage compared to old-fashioned studying and comparing? Maybe I'm missing something?

You are able to see and compare details that are almost impossible to differentiate when scanning from one photo to another, particularly if you toggle layers on and off.

Originally Posted by zapokee

Very nice idea.

Knowing how much variation there was with Rolex/Tudor consistency back in the 60s and 70s, I wonder how concerning it should be if overlaid images don't match up exactly.

The printing even on very old dials, is consistent and all I can go publicly.

[interestedwatcher]

The most difficult thing I am finding is getting high enough resolution images to accurately compare. Whilst zoomed a non perfect image can make specific letters look different, when in fact they are the same but the image sucks..for example 0, a poor image can make the inside of the 0 look bolder that it is, so if you compare to a high resolution image, that same 0 looks ever so slightly skinnier.

I have found pretty universally that none of the fakes so far have been able to line up the Coronet, the sub Coronet text, the depth rating and the swiss or T swiss T. If you can overlay your dial and line all of these things up with a few pixels here and there, it's looking good.

[interestedwatcher]

Originally Posted by linesiders

You are able to see and compare details that are almost impossible to differentiate when scanning from one photo to another, particularly if you toggle layers on and off.



The printing even on very old dials, is consistent and all I can go publicly.

Another good option is to create GIFs, this is 11 images from 0 percent opacity to 100 stepped at 0.3 seconds.

[interestedwatcher]

This is a point when comparing dials etc online, resolution is everything.

The below picture is from the same dial, one of the images is a little over exposed, as a result under extreme zoom (500 percent), the image is not so good. Accordingly, this looks like two different fonts, but it isn't. It is the same dial.

This will happen when you have a VERY high quality image that is being overlaid onto a not so good image.



So what becomes very important for me, is the almost exact (allowing for slight variations if the image is taken from a 1 - 2 degree different angle) alignment of everything else on the dial. The coronet, the lume plots, the ROLEX, and the depth markings, if you've lined a couple of points up perfectly (I generally work off the coronet, and the outer track if it has one), then everything else falls into place.

Example



and

[interestedwatcher]

Did a comparison for someone of a OCC service dial. Obviously different dials.



Note - doesn't mean the comparison dial is fake, just that what it is being compared to isn't the same dial.

If anyone wants a comparision, PM me with what you want compared to what, and I'll do it.

[joli160]

Nicely done

[Styles Bitchley]

Very creative idea.

[interestedwatcher]

Comparison on the dial back of a couple of Snowflake dials. One is a bad redial, but both are on original base plates.