Released 22/03/2024 Intro video: Zo0lon Video Edit: Zo0lon Intro Music: LukHash Vote/Download: csdb.dk/releas... - The Ghost csdb.dk/releas... - No Sprites
Some brilliant insights shared here, that many (like me) would surely have overlooked or missed otherwise. You guys continue to push the boundaries after all these years, which is truly inspiring. Congratulations on the well earned victories!
1:46:46 some clever timing on the movement of the (xFFF) + scroll-register (D016) going on there. I guess it's table lookup, since the movement is consistent outside the border + when it goes into the border. Well executed. Suspect there was some tweaking of the sinus data to avoid mishmash. Need to look at the code to tell for sure, but reminds me of the 9 sprites in the border we used in Panoramic's Mentallic demo, the last part (Olav + Bjørn) came up with the first movable xFFF split timing for the 9th, which had to switch between two of the letters, as they moved they became sprites, but the 9th had to switch place and timing on the exact cycle when movement starts. EDIT: surprised! you included it - 😂
@@FairLight1337 Old scener. Been into demo making since 1986. I lived close to Razor 1911 so that helped. Mailtraded for many years with other groups, like D-USA Team, S451, HZ, 711, Censor, Triad, FLT... I enjoyed getting demos more than games. Been watching your channel regularly. Good stuff... 👌👍
Could we get two things, please? 1) a part two of this interview where we go further in depth (also on the demo ‘eyes’) 2) an interview with fegolhuzz, who is hands down the best SID-er at the moment That would be so amazing!
almost on the contrary - you guys speak about the ghost byte(s) and the raster/cycle timing etc. as if it's common knowledge. It would be very interesting to understand more of it and why it is such a feat/challenge. Actual code would go over my head, but understanding "there are X cycles and the bit of code to run this needs Y and therefore timing here and there is crucial' (if that makes sense) would enhance my (and I suspect others') appreciation of what's really on display here. So maybe more about the limitations and how Trident manages to figure out how to get around them. @@FairLight1337
actually, I just found the sessions on Fjalldata 2023/2024 where Trident explains some of the special effects seen in recent demos and that's exactly what I was looking for. So, if you ever get to it, something like that 🙂@@FairLight1337
I already thought the name sounded familiar, used the IP stack and protothread framework a lot on 8051 style microcontrollers, Duff’s device is an absolutely brilliant trick :) Really cool interview!
I love it as well. As mentioned, I wrote the generally same routine for Swedish computer Magazin Datormagazin in the 90s, but Adam added a lot of stuff to it that makes his variant so much better.
Kontiki was a Norwegian 80s Computer, it got renamed to Tiki later due to Heyerdahl didn't approve Tandberg used his name. Tiki 100 was our school machine. There has been released demos for it (Solskogen) showing what it is capable of. Not bad for it's time.
Oh come on (dore) GPT just putting words in order ? The reason why GPT is not good at generating C64 code is that it was trained on doc only, not trained to produce code especially. Like you would not speak great in Intalian, whit just the help of a dictionnary. Same thing here, GPT needs italian courses. Working on it ;-)
@@FairLight1337 to my knowledge, there is no model trained on C64 code base yet. I'm working on it. I'm teaching AI, and working on building a dataset of C64 docs (form books, newspapers, etc) through the courses. I will adress source code soon. Right now I can ask questions that will help a developper (what's the adress of.. how many cycles for.. etc.) but not good at generating code which is normal