Just bought this from the same seller that I got the "scratched" CACG coin from. Same terminal die state. Given the same seller and state, I suspect these are from the same roll(s). I'll be stoked to get this one in the mail...
My prior measurements of rotational face offset were off. I am using an algorithm to measure -- it works well doing one certificate at a time, but it breaks when I try to run more than three at a time. I can't quite figure out why.
I burned through a lot of tokens and Grok credits trying to get AI fix it - it's better but it's still not precise enough for my purposes. If anyone has tips or techniques to automate measurement of rotational die face offset of coins already slabbed, please share.
@Maxvan1 said:
My prior measurements of rotational face offset were off. I am using an algorithm to measure -- it works well doing one certificate at a time, but it breaks when I try to run more than three at a time. I can't quite figure out why.
I burned through a lot of tokens and Grok credits trying to get AI fix it - it's better but it's still not precise enough for my purposes. If anyone has tips or techniques to automate measurement of rotational die face offset of coins already slabbed, please share.
Think I found a better way – I don't need the actual rotation, I just need the rotation relative to one and other.
Using one a base layer reference image with transparent overlay of the target certificates works much better. The bot just has to rotate the target layers until the coin images match the reference layer... It's more like a captcha ("slide the piece"), which the robots are much better at.
Okay now that we've got the rotational order right, it's starting to make more sense...
The images below are in rotational order (relative to one and other). You can see the relief is sharpest at the two extremes, and mushed toward the middle of the rotation. This suggests a slight cock in the tooling ("die cock" or "DC" error). Definitely want to expand the set to include 40 coins to confirm.
The observed rotations in this grouping of 18 suggest the die walks. Sharp relief at the two ends of the rotational walk suggests the DC results from the tooling. When the Mint Workers reseat the die, they flip the tooling over which then changes the side that is DCed. In "Seating A" it is cocked toward 9:00ish and rotating CW; in "Seating B" is it cocked toward 3:00ish and rotating CCW.
The DC is most evident in the Y on the obverse and the lower hem of the Eagle's left (viewers left) wing. In "Seating A" the metal is pushed toward 9:00ish, in "Seating B" the metal is pushed toward 3:00ish. Where the two seats overlap (the center of the rotational range) is smashed / strongest DC. The rotation and direction of the DC also determines which cracks are visible on Washington's head, which provides a hypothesis to answer @IkesT's question how the die cracks can appear so differently - it depends on the whether it's in "Seat A" or "Seat B" (i.e. DC direction) and the degree of rotation. Need to expand the sample set to confirm, but that's the current hypothesis.
"Stage E" of the die progression is after a heavy lapping. The heavy lapping process that takes out the Hicks leaf on the 1956 appears to be the same lapping process that takes out the arrow tips on the 1962/1963.
This is gunna be such a sick progression set when it's done...
*Edit: Marked Stages are incorrect and inconsistent on these plates. Need to be updated.
Comments
Not a gouge.
Just bought this from the same seller that I got the "scratched" CACG coin from. Same terminal die state. Given the same seller and state, I suspect these are from the same roll(s). I'll be stoked to get this one in the mail...
Terminal state = indicator crack enters wing.
My prior measurements of rotational face offset were off. I am using an algorithm to measure -- it works well doing one certificate at a time, but it breaks when I try to run more than three at a time. I can't quite figure out why.
I burned through a lot of tokens and Grok credits trying to get AI fix it - it's better but it's still not precise enough for my purposes. If anyone has tips or techniques to automate measurement of rotational die face offset of coins already slabbed, please share.
Even better mark> @IkesT said:
Yeah I agree. See revised description and markers. Thoughts? Not pareidolia.
Think I found a better way – I don't need the actual rotation, I just need the rotation relative to one and other.
Using one a base layer reference image with transparent overlay of the target certificates works much better. The bot just has to rotate the target layers until the coin images match the reference layer... It's more like a captcha ("slide the piece"), which the robots are much better at.
Okay now that we've got the rotational order right, it's starting to make more sense...
The images below are in rotational order (relative to one and other). You can see the relief is sharpest at the two extremes, and mushed toward the middle of the rotation. This suggests a slight cock in the tooling ("die cock" or "DC" error). Definitely want to expand the set to include 40 coins to confirm.
The observed rotations in this grouping of 18 suggest the die walks. Sharp relief at the two ends of the rotational walk suggests the DC results from the tooling. When the Mint Workers reseat the die, they flip the tooling over which then changes the side that is DCed. In "Seating A" it is cocked toward 9:00ish and rotating CW; in "Seating B" is it cocked toward 3:00ish and rotating CCW.
The DC is most evident in the Y on the obverse and the lower hem of the Eagle's left (viewers left) wing. In "Seating A" the metal is pushed toward 9:00ish, in "Seating B" the metal is pushed toward 3:00ish. Where the two seats overlap (the center of the rotational range) is smashed / strongest DC. The rotation and direction of the DC also determines which cracks are visible on Washington's head, which provides a hypothesis to answer @IkesT's question how the die cracks can appear so differently - it depends on the whether it's in "Seat A" or "Seat B" (i.e. DC direction) and the degree of rotation. Need to expand the sample set to confirm, but that's the current hypothesis.
"Stage E" of the die progression is after a heavy lapping. The heavy lapping process that takes out the Hicks leaf on the 1956 appears to be the same lapping process that takes out the arrow tips on the 1962/1963.
This is gunna be such a sick progression set when it's done...
*Edit: Marked Stages are incorrect and inconsistent on these plates. Need to be updated.