Kaleun96

I've never photographed a gold coin so can't add much specifics to this discussion but I can highly recommend the book Light Science & Magic: An Introduction to Photographic Lighting to help with problems like this. It touches on many aspects of lighting for a range of materials and desired outcomes without getting overly technical or going into too much detail. The examples are well documented with example photos and diagrams so I find it easy to jump to a specific section of the book without having to read everything that came before it. For this situation the advice would probably be similar to what has been said already but it'll help you hone in on achieving that specific lighting, e.g. what range of angles your light needs to be at relative to the coin and give you some tricks that you can try to balance the ratio of direct vs diffused reflections. Concave coins are definitely tricky, it's hard to balance getting even lighting across the whole surface without making the coin appear too "flat". So I would probably use both high-angle non-diffused lighting with a mix of diffused lighting that can reach all the parts of the coin. It shouldn't matter much whether you're using flash or LEDs or sunlight, the control of brightspots/over-exposure is entirely up to you, it's not ingrained in the light source. Similarly with the background, the colour matters little if you have control over your camera. The direct light will be your "key" light, i.e. the one providing the main source of reflections and is contributing the most to the look of the resulting image. With a concave coin, it will likely make some parts of the concave edge opposite the light look different to the edges closer to the light, simply due to the different angles relative to the light (one way to avoid this is with axial lighting). So you would then use your diffused light, the "fill" light, to help balance that out so there's not such a big contrast in the exposure of the edges but you don't want the diffused light overpowering the image and filling in all the shadows. Oh another thing you can do to help with getting even exposure from the concave edges is to tilt the coin towards the "key" light so that the surface of the coin is no longer perpendicular with the camera. This might require focus stacking due to the increased "depth" of the coin but it's what some auction houses use to help get high angles of reflection without using axial lighting (for example, see this commercial setup that some auction houses use). edit: I see a free PDF version is available online hosted on someone's website but I'm not sure if this is being shared legally so won't link it directly here. It was the third result on Google for me. I'll reproduce one of the figures from it below to illustrate what the book is like for those interested. This particular figure explains how the angle of the light relative to the subject doesn't necessarily determine whether the reflections will be direct/specular or diffused, instead the distance from the light to the subject needs to be considered to know what reflections you will ultimately get.

A friend knew someone with a copy of the Elder plates and was able to get some better photos for me, including of the reverse this time, to confirm the match beyond all doubt:

Classical Numismatics was kind enough to make a video on Alexander tetradrachms based on a few videos I had taken for my 2023 Top 10 list. I love being able to share part of my collection this way and it's great how Classical Numismatics is so often building his video essays around coins that everyday collectors like us have rather than just those in museums or feature auctions. He also managed to tie-in a few of my other coins that are tangentially related, like the Alexander-type tetradrachms in the name of Philip III and Seleukos I, as well as a Philip II tetradrachm. There's one correction to make for the coin at 7:00, this one is actually from Kalchedon and is a posthumous type. The coin featured at 12:00 was one of my favourite pick-ups of 2023, not only because it's a lovely example of a tetradrachm from Amphipolis but also because it's ex. Edward Newell, one of the most important numismatists of the 20th century. Though up until last week, I could only go on the auction house's word that it was ex. Newell, who were in turn relying on a Spinks auction from 1982. The best theory was that the coin was sold in the Parke-Bernet auction of 1968 which auctioned off some of Newell's estate that didn't go to the ANS, after which it was sold at Spinks and then at Noonans last year. But a few days ago I spotted it in the one plate from Thomas Elder's "Remarkable Collection of Greek Tetradrachms" that is available online, where it is coin #4 from Plate I. Thomas Elder had acquired something like 1,000 tetradrachms from the Demanhur hoard shortly after it was discovered and brought them back to the US. He then published 6 plates cataloguing 300 of these coins circa 1910 and Edward Newell then bought many of these coins from Elder, as he says so himself in Demanhur: So not only is this coin all but confirmed to be ex. Newell, it can also be said to be from the Demanhur hoard (which was suspected anyway), ex. Elder's collection, and also plated in Elder's work.

I don't see what the problem is, just post the coin and share your opinions and everyone can make up their own mind. If the auction house has rebutted some of your arguments, it'd be of course fair to mention those as well seeing as it's unlikely they'd comment here themselves. As long as you do it respectfully and fairly, there should be no issue, especially after you've already contacted the auction house privately first to give them an opportunity to weigh further evidence. Since there still remains a disagreement, I would just post something along the lines of "The auction house says it's from this mint for x and y reasons, I disagree and think it's from this mint for this and that reason". No harm, no foul. We're allowed to publicly disagree with auction houses, especially for types that lack a robust attribution. Academia doesn't pussyfoot around like this, no reason we must either and I bet we can be a lot less snarky while we do it compared to many academics.

"Unlabeled AI tools are unlikely to work ... Coin images will need some labelling, starting from 'this is an ancient coin'" You of course need to do some data cleaning beforehand, as you do with any kind of die study, but this is not what is meant by "labelling". Labelling would be identifying the dies in advance and validating the performance against a test set. When we're talking about training an unsupervised model, we mean feeding it images of coins you want it to cluster together based on similarity. So there's no step where you need to tell the model "this is an ancient coin" unless you plan on feeding it images of bananas or cats. " And here we have 100k+ ancient coin types for labelling!" Perhaps there's a misunderstanding. From your reply I get the impression that you're thinking of some generic AI model that is capable of producing die studies for any coin you feed it. That's not what I have in mind nor is what most people have in mind. Maybe one day it will be possible to just feed a massive set of images to a model and tell it to identify all the dies but that's neither necessary for ML/AI die identification to be useful nor the intended use case currently. What the current models are proposing to help with is the following scenario: you have 10,000 images of coins of a particular type or series that you want to do a die study for. You scraped/downloaded these images from some database that contained other information, such as the ruler and possibly type attribution. You've gone through and cleaned out the irrelevant results, tidied up some of the attributions, and now you have 9,700 coins to perform a die study on. The next step is either to do a manual die study or feed these images to a model. You've spent a long time cleaning the data but the die study itself is going to take even more time. This is the point the model steps in. You'll want to train the model on specific types or groups of related types (i.e. similar obv/rev designs). But this step is not some additional step that only an ML model requires - you have to identify the types in advance for any kind of die study. So in short: no one is proposing to feed an ML model a trove of random coin images. The data cleaning and organising aspect of die studies isn't going anywhere anytime soon. You still need to know some basics like the dataset contains coins of the same emperor/ruler/iconography and are either the same type or multiple similar types. That data cleaning and processing does take awhile but the actual identification of dies takes a significant amount of time too and it's this specific part that supervised or unsupervised ML models can help with. That being said, you can absolutely have fakes/duplicates/misattributions in the dataset and not cause any significant problems - these will be easy to identify and remove once the model attempts to cluster them or will be part of the general post-clustering validation. You can find an example of an unsupervised model for die identification here: https://www.researchgate.net/publication/356710176_Unsupervised_Statistical_Learning_for_Die_Analysis_in_Ancient_Numismatics Once a die list is known for each coin type, like some Syracuse masterpieces, do we need AI to link a new coin? I don't exactly get your line of reasoning here. There's not a dichotomy where we have to commit to either only using AI or never using AI. If all the possible dies are known then whether you need/want to use AI to identify the die of a new coin is entirely up to you. It may help for types with a large number of dies, and it'll definitely help when processing a new hoard even if there are no new dies in the hoard. Picture a hoard of 30,000 Athenian tetradrachms - would you prefer to go through the list of, say, ~1000 known dies and identify the correct die for each one, or would you rather a model do that work for you?

Since you said AI is unlikely to help currently and then listed problems with die studies that would also apply to AI models, I assumed that those were reasons why you think AI models would not help currently. But if those are just general reasons, what are the reasons for why you think AI models specifically aren't yet useful? I think these models are an emerging technology when it comes to die studies but I don't see any inherent limitations that mean they aren't yet useful. The main problem is that most people capable of a die study aren't statisticians or proficient in coding and we're probably a little ways off from a generic model being useful for most coin types, meaning bespoke models trained or optimised for different sets of data is still required. On a dataset of ~450 coins, I've probably spent half the time invested just identifying dies and as mentioned the time to check one die increases the more examples you add. A model could check a new entry's dies against the rest of the database in milliseconds, and at the least probably rule out the vast majority of possible matches and leaving just a couple close matches. The only reason I haven't tried to make a model myself is that it requires upfront investment to learn that aspect of machine learning and I've already done the bulk of the die matching. It's been in the back of my mind to try one day, maybe when I start a new die study from scratch.

You shouldn't need to tell the model what to focus on per se, it can work that out itself. There are tweaks you can make so it doesn't focus on the wrong thing or you can add weights to certain features etc, depending on what type of model you employ, but theoretically you wouldn't need to be that familiar with the coinage. Being familiar would help with the validation and optimisation of the model though. You may need some sort of training dataset to validate the model on and that would require manually die matching some examples and then checking accuracy but it's probably possible to do this completely unsupervised too. You feed the model the images and it learns the differences, you then make tweaks to make sure it's identifying unique dies rather than examples, and then manually check its results after. I don't know whether supervised vs unsupervised models are more common for die matching but I don't see why both wouldn't be possible.

A lot of those points would apply to manual die analysis using only photos too, which a lot of die studies these days rely on. But I also disagree with some of them. For example, getting records/images can be as fast as a script that scrapes a database or polls an API. If you're recording them by hand then yeah it will take a long time but you can also download 10k records in a matter of minutes depending on the source and data quality. The problems you mention like multiple photos of the same coin, poor image quality, or tooled coins do mean manual validation or data processing is always required or at least recommended. But a die matching model could also theoretically flag identical coins and once the model has identified coins by their dies, it's much easier to manually filter out duplicates or invalid results. These are problems I've faced when die matching manually from photos and having a model go through first and do some clustering would speed up the process significantly. I think these models are absolutely the future and will unlock die studies previously deemed impossible/too tedious. Manual validation will remain part of the process but 90% of the work could be done by models in a fraction of the time a human can do it and then the remaining 10% is pre-cleaning the data and validation.

I think you have to be a bit careful drawing any conclusions so early into the year and also from specific auction houses/dealers. January is a feature auction month and e-sales tend to be a bit weaker following them. We all know what's going on with Roma. Re: Leu, a lot of their inventory appears to come from hoards so you see some big swings in trends for certain types of coins. But excluding hoards, large consignments can dominate an auction house's calendar for half a year or more and we may never know it if the consignors choose to not use a collection name. An example of this for my own collection area is the Don T. Hayes collection that CNG has been selling off but his collection included a lot more than just Alexander tetradrachms. CNG started selling them in March 2023 and it's still going, though it feels like it's dying down a bit Certain dealers also often get a sudden influx either from a consignment or hoard. Regular collectors also might be less likely to have consigned their coins in Nov/Dec last year than at other times during the year So I think it's just the normal ebb and flow. Sometimes I'll go months without buying a coin from CNG or Leu just due to the kind of coins they're consigning at that time. I just assume that while those sales may not be interesting for me, I'm sure they're interesting for someone else. Those other people are probably glad that finally these auction houses are back to selling coins they collect, while for me those previous sales were perfect for me.

This is just what the DHL rep told me, they'd not cover the loss of any goods considered to be antiques, which included collectible coins older than 100 years. I can only find reference to this for DHL Freight and DHL eCommerce services. I can't see any terms and conditions for DHL Express that explicitly exclude certain goods from the insurance for that service but I also didn't look for too long.

Hopefully I'm not derailing the thread too much from the intended purpose but I have always wondered how auction houses offer insurance on shipping. From my understanding, both DHL and FedEx refuse to insure ancient coins (at least when I have asked), as well as my local national postal service. So when I sell coins I always explicitly state that they're not insured even if the shipping option I used says insurance included. I believe some auction houses have insurance provided through a third-party rather than the carrier. It's possible this auction house is doing the same and hence the arbitrary doubling of the price and lack of insurance on a FedEx parcel, or they're offering their own "insurance" by working out how much $$$ they lose in lost packages each year and how much they would have to charge for shipping to recoup that expected loss.

I think it's likely pitting from corrosion rather than casting bubbles too. It's not too uncommon to see on a lot of the low budget Alexander drachms we see coming to market on biddr. In terms of Zurqieh, I wouldn't say they're unlikely to sell fakes as they absolutely have (and removed them when alerted) but they're probably better than your average budget biddr auction house. They have the help of a reputable numismatist for checking attributions and authenticity I believe but even one fake that I know of has slipped past him and it was a known forgery.

Congrats on nearly getting the map and timeline done! Always cool being able to visualise your collection in different ways. Re: the timeline chart, as you mention it's not ideal for some devices yet due to the page width. One thing that helped me with scaling my own timeline chart on narrow-screen devices was putting the chart in its own container with overflow-x set to auto, that way the page itself doesn't stretch in width but the user can still easily scroll across the width of the chart 🙂

50% unsold at this point sounds about right for the last few Buy or Bids from HJB. I think you have to go back two years or more before you find the Buy or Bids that used to see huge numbers sold in the first few hours. Every time you would refresh the page, another couple of coins would be sold. It was a mad rush to go through the catalogue before all the good coins were sold, only hindered by their slow servers (which have gotten better) and buggy website (which is about the same). My run of not buying anything from a Buy or Bid after is nearly up to 3 years now, and that was after I bought five or six coins in them between 2020-2021.

It's always a possibility but I think its likelihood depends on the type of coinage - do you have 31 examples of the same Alexander III tetradrachm by any chance? The frequency of dies, die pairings, types, examples, availability in public collections, and prior research affect how likely it is for someone to amass such a collection so what makes sense in one field of ancient numismatics may not in another. If it were a die study, we likely would've seen that die study published before they sold the coins and we'd expect to see provenances. Perhaps they did it for personal reasons (or didn't finish it), you never know, as I haven't come across such a study myself and I've also yet to see a recent (i.e. not Newell etc) die study of Alexander III tetradrachms that came from the author's large personal collection. All the ones that come to mind rely on a mix of public collections, hoards, and auction results. Additionally, this "collection" doesn't represent every type from Aspendos, Phaselis, and Perge and I bet it doesn't represent every die of the types it does have either. So it would seem to me odd to have four of one type, including of the same die, but none of others. It just doesn't strike me as a personal collection at all, unless it came from someone who bought it as a hoard themselves.