Finally getting back into some game dev after a busy weekend. My claude usage resets tonight so time to spin up some new projects to burn some tokens. Tomorrow I hope to return to Trenzelor and get it wrapped up with a simple loop for Itch.
I plan on testing locally with images only on my 3090 but I wanted to know if you had any clue on these questions, If you have the time I'd really appreciate it I am just so out of my depth and super specifics tend to not get great answers from claude etc.
1. If I needed to use your online service to train a model, maybe a show model so I don't need to use references for my cast (and to get better vocal performances) what would you estimate the cost to end up being? I noticed the jacked lora ended up being only like $13 (or at least your credits went down that much). Now obviously thats a simple concept while a show with lots of characters may end up being a bit more. But I have no clue what kind of scale to expect with all this stuff.
2. Is it better to train character or full cast LORAs?
3. How trainable is just audio? I get pretty poor dialogue most of the time so if I could basically train the performance of a character I feel like that would be helpful but I'm kinda unaware if that's even a thing.
4. You said 3090 on images training would probably be possible with full offloading but idk how much offloading that would require. I only have 32 gb of ram right now so I guess in theory I would maybe need to upgrade ram (or use that money training lora's I guess lol)
Again sorry for the long questions and such I just need some direction to head in and I'm a bit lost right now while I'm deep in production into so many active projects.
Swapped to single character + multiple dice. Do you think its better to have a party with single dice per or the multi dice and the ability to rescue more party members on a run?
#screenshotsaturday#indiedev#gamedev
Testing out elite fight tiles that are deeper than the standard dice roll off and actually let you do DND-ish fights. It was how I was doing the final boss but I felt like it was a waste to only do it once per run. Also I'm considering dropping it down to be a single character per run with the possibility of recruiting other characters during (also maybe creating specific runs/difficulties that are built for bigger parties).
Regardless I've played a few runs now and it's a ton of fun, really scratches that TTRPG itch. After I get the push-your-luck dungeon crawl version ironed out I definitely plan on building out a full RPG system with the basics I already have and the other systems I built and didn't end up using like inventories etc. So if either version sounds interesting give me a follow I'd really appreciate it.
#gamedev #indiedev #TTRPG
Added obligatory rogue-lite upgrades for the modern gamer :) Pretty much placeholder tier design for the time being but I feel the game leaving tech demo tier and reaching into something playable.
#gamedev#indiedev
@seraphsanctity For years all my stuff just sat unpublished so I guess forcing myself to be rigid is at least forcing me to get stuff out there. Definitely times I feel like there are some big sacrifices in the projects whether its quality or vision but at least others are getting to see it.
Episode 1 if you missed it. It's a mockumentary that parodies MTV/WWE/Influencers/Food Shows/and more, but in medieval times. It's a lot of fun!
https://t.co/pMVHJhF4O5
Testing a handheld resolution so I can get this put on my Anbernic RG35XX H after it's done, the main PC version will have much better info representation but I'm impressed with what I was able to have readable on such a small canvas. What do you think?
Some cool Trenzelor progress. Got a zoom/scale effect to show you being at a different height, needs a bit of polish but imo adds a ton while keeping the kind of tactile graph paper feeling. Art being ported in for the characters is next!