Post by dshadoff on Jan 21, 2019 16:54:25 GMT
I just extracted all of my HuCard data over the past few days, now that I finally have a TurboEverdrive and the UperGrafx is also getting SDCard instant-load functionality.
My old images were all from the '90s - many of them were extracted my own copies on the floppy-disk-based copiers of the day, but since that was tedious, I had obtained other images from other sources in order to save time and effort.
As I performed my re-extract, I did binary compares - I had removed the headers from the old ones, since the UperGrafx images are all headerless. These compares showed me a variety of differences - some of which I expected, and some of which I didn't.
Type 1: Dirty contacts.
This showed up as a wrong ROM length, or large numbers of differences in (usually) regular patterns. The solution is first to clean the HuCard, and see whether the results are any different. This actually showed up around 15% of the time. (I used an eraser to clean the contacts... works more or less about as well as alcohol).
Type 2: Bad original image
If I got the same read a second time, I went looking for an alternate "second source" image. I ended up doing this for 20 ROMs out of the roughly 300 HuCards... which is actually not a small numbers. In 14 of those cases, I found that my original image was bad, and the new image agreed with my extract.
Type 3: Deliberately altered image
There were a small number (say 3 or 4) which I didn't try to re-source - these were identical matches, except for obvious, explainable differences. In at least one of these cases, the only difference was that the copyright message had been changed to spaces. (I didn't keep details on these.)
Type 4: Unexplainable
In about 7 cases, the second-source image was the same as my original image from the '90s, but differed from the new extract.
Two of these (BeBall and Darius Plus) differed by only one byte, and that byte appeared to be a hand-made change, but it wasn't clear why it was made.
R-Type 2 looked like a deliberate but small change in a section of assembly-language.
...But I had 4 which were large enough to question:
F-1 Triple Battle
PC Genjin 2
Power League 4 (fairly large differences, but they appeared as though they might be in a statistics block)
Street Fighter 2
...So I started to wonder whether the bad images were the ones on the internet, or whether my cards had decayed.
Since I saw a greater-than-expected number of dirty-contact bad rips on my own cards, it's entirely conceivable that some of the rips on the internet were also bad - and never got replaced, propagating endlessly.
I am going to have to obtain new copies of these HuCards before reaching a conclusion - I'll only be happy when two independent extracts match. Luckily, I don't think any of these cards will be too hard to find next time I go to Japan (and possibly even before then).
Dave
P.S. Another odd finding was that "TV Sports Hockey" (Japanese release) is a 6 megabit game (and also a 6-megabit image everywhere), but is actually delivered on an 8-megabit HuCard. The last 2 megabits are all zeroes; if it was actually a 6-megabit HuCard, the last 2 megabits would be an "echo" of one of the other banks, or worst case, all 0xFF's (not 0x00).
My old images were all from the '90s - many of them were extracted my own copies on the floppy-disk-based copiers of the day, but since that was tedious, I had obtained other images from other sources in order to save time and effort.
As I performed my re-extract, I did binary compares - I had removed the headers from the old ones, since the UperGrafx images are all headerless. These compares showed me a variety of differences - some of which I expected, and some of which I didn't.
Type 1: Dirty contacts.
This showed up as a wrong ROM length, or large numbers of differences in (usually) regular patterns. The solution is first to clean the HuCard, and see whether the results are any different. This actually showed up around 15% of the time. (I used an eraser to clean the contacts... works more or less about as well as alcohol).
Type 2: Bad original image
If I got the same read a second time, I went looking for an alternate "second source" image. I ended up doing this for 20 ROMs out of the roughly 300 HuCards... which is actually not a small numbers. In 14 of those cases, I found that my original image was bad, and the new image agreed with my extract.
Type 3: Deliberately altered image
There were a small number (say 3 or 4) which I didn't try to re-source - these were identical matches, except for obvious, explainable differences. In at least one of these cases, the only difference was that the copyright message had been changed to spaces. (I didn't keep details on these.)
Type 4: Unexplainable
In about 7 cases, the second-source image was the same as my original image from the '90s, but differed from the new extract.
Two of these (BeBall and Darius Plus) differed by only one byte, and that byte appeared to be a hand-made change, but it wasn't clear why it was made.
R-Type 2 looked like a deliberate but small change in a section of assembly-language.
...But I had 4 which were large enough to question:
F-1 Triple Battle
PC Genjin 2
Power League 4 (fairly large differences, but they appeared as though they might be in a statistics block)
Street Fighter 2
...So I started to wonder whether the bad images were the ones on the internet, or whether my cards had decayed.
Since I saw a greater-than-expected number of dirty-contact bad rips on my own cards, it's entirely conceivable that some of the rips on the internet were also bad - and never got replaced, propagating endlessly.
I am going to have to obtain new copies of these HuCards before reaching a conclusion - I'll only be happy when two independent extracts match. Luckily, I don't think any of these cards will be too hard to find next time I go to Japan (and possibly even before then).
Dave
P.S. Another odd finding was that "TV Sports Hockey" (Japanese release) is a 6 megabit game (and also a 6-megabit image everywhere), but is actually delivered on an 8-megabit HuCard. The last 2 megabits are all zeroes; if it was actually a 6-megabit HuCard, the last 2 megabits would be an "echo" of one of the other banks, or worst case, all 0xFF's (not 0x00).