Jump to content

sanahtlig

Backer
  • Posts

    3425
  • Joined

  • Last visited

  • Days Won

    33

Everything posted by sanahtlig

  1. The quality and variety of sprites will probably matter more than the number of CG in this sort of game. Reserve full CG for climactic points that you want to highlight. But I suspect your question is actually about the game's presentation on your store page. CG make for good promotional images. Your Steam page is definitely suffering from a shortage of high-quality screenshots--you even included a fairly non-descript title screen! Text screenshots can be ok, but the re-use of the same background in all such images gives a cheap/bland feel and evokes a scope that is suffocatingly claustrophobic (are we really going to be talking to the demon on a single static background the whole game???). Right now the most evocative screenshot is probably the one of the puzzle gameplay. You need variety and movement in the artwork (sprites, background, CG), in the game also but especially in the screenshots. Even if the game takes place entirely within one room, you'll need multiple backgrounds and perhaps versions of those backgrounds. If your focus is on horror, simple filters over the backgrounds (blur, change in color, etc.) at appropriate moments could add a lot to the atmosphere (really key in the horror genre) without running up costs. And if push comes to shove, generative AI is always an option for increasing variety at low cost--it seems to be pretty decent at backgrounds. In short, by making good use of sprites and backgrounds, you'll actually reduce your dependence on expensive CG to create movement and immersion. I suppose I should give a brief nod to what I like too. The character artwork is appealing and the single background you've shown is adequate for the task. The quality is there, you just need sufficient quantity. The screenshots show multiple high-quality sprite poses, which leaves a favorable impression, though showcasing changes in facial expression would further increase appeal.
  2. For those with no JP knowledge, it's probably best to use a combination of grammar and machine learning based models. For those with the ability to check the TL accuracy, the machine learning model could be sufficient. The processing power problem can be solved with GPU-processing. GPUs are more than capable of handling the load. The problem with grammar-based MTL is that it's not very accessible to the typical user. It's less likely to mislead you, but the botched syntax takes more effort to interpret. Most users prefer a readable TL with camouflaged mistakes over one written in a pseudo-English that has to be painstakingly deciphered.
  3. There is no version with offline DeepL. DeepL requires an online connection. The latest release can usually be found on his Patreon page.
  4. The Offline Model V4.0 is dated Dec 2022. As far as I know, It's not the default in the current package (Sugoi Translation Toolkit 4.0), so it takes some sleuthing to find and install.
  5. I'd recommend Sugoi Translator with the updated model installed. It runs locally, with output as good as or better than DeepL. You can read my more detailed write-up on the tool here.
  6. That would imply a commitment they're not actually making. The idea behind crowdfunding is that you're funding a cause that's important to you, important enough to take a risk on. If you're an eroge fan backing an eroge localization project, it's kind of fundamental that the end product be an eroge. You can bet that the suits at corporate don't give a damn about making good on the vague suggestions some random staffer is making in an obscure Discord channel to a handful of Western fans across the globe--especially if they run into sexual themes issues with the Steam release. And if the patch that "appears" has technical or compatibility issues? They won't care about that either. Trust me: been there, done that.
  7. That message uses terms/slang I'm not familiar with. Is スペコミュ communicating [a game's] specs? エイベックス seems like the name of an entity, perhaps of a company or the pseudonym for a representative. Seems like he's fretting that overseas fans don't seem content with vague suggestions that funding the Kickstarter non-adult project might someday result in an adult release. Which, well, seems like a perfectly reasonable stance. Kickstarter projects often don't fulfill their stated commitments, much less their unstated ones. If they want buy-in from the eroge community, they're going to need to show a concrete commitment. They could always take a page from Ninetail's book: partner with an eroge publisher who will take responsibility for the adult release. After the ridiculousness between Sekai Project and GIGA, with GIGA appearing to actively block an adult release, it's hard to give JP companies the benefit of the doubt anymore. The problem is not that nuance isn't being communicated across the language barrier. It's them not showing commitment on a topic that's important to their Western fans, many of whom have gotten burned on this before.
  8. I use Translation Aggregator for MeCab and JParser. You're probably better off using another visualization tool if you want to use machine translation. Textractor has a variety of plugins, and Sugoi Translator has its own deep learning model that's trained specifically for VNs.
  9. Any games with sexual themes involving high school students are subject to arbitrary banning on Steam. The anime adaptation, while not a hentai, is raunchy enough that it would qualify. To avoid this risk, they'd have to gut interactions with one of the main heroines in the opening segment, plus remove multiple routes involving younger characters. The Tweet is announcing a Discord server for the crowdfunding campaign. It would perhaps be more useful to quote the actual announcements (from the Discord server?) rather than the tweet. Did they mention a date for the Steam release?
  10. I looked at the Steam page but I still feel like I have no idea what the game is about. Let me explain my thought process to maybe help you refine the presentation. First I see this: This is extremely vague. Two vulnerable individuals? The title picture has two girls in it. Is this yuri? Next I look at the screenshots. There's no text. My first-pass judgement is that you're hiding the text for some reason. Is it machine-translated? (I've been conditioned to ignore Steam videos because they're often not informative, and only later did I realize the videos show gameplay with text.) Finally I get to the full description. Here I finally learn that we have a male protagonist. But then I'm slammed with several confusing messages. "He is about to graduate from Tokyo University of the Arts". So he's 22? But "he has four wonderful girls". Is that a roundabout way of saying daughters? So he's not 22? Or did he start getting girls pregnant at the age of 12? Why is he studying art with very little prospects for a stable income now or in the future if he's a single parent of 4 daughters? "His alma mater, Tokyo University of Arts". Wait, I thought he hadn't graduated yet??? This was supposed to be a love story between two characters. Who is even the female heroine? Is it one of his 10-year-old daughters?! At this point I somehow have less idea what this game is about than when I started. You need to work on the presentation if you want to get first-time viewers who happen onto the Steam page to not hit the back button.
  11. A simple question during signup like "What is the name of this site?" or "What is our slogan?" might be effective for deterring spam.
  12. But what's the point of referencing French revolutionary history if it's just going to be another slice-of-life school setting...? Where's the chaos, cult of personality, grandiose dreams of empire, bold military maneuvers and conquest?
  13. Wow, Softhouse Chara and then GIGA. The titans of gameplay eroge are falling one-by-one.
  14. Is Gore Screaming Show a moege? Discuss.
  15. Do Western users actually play VNs on consoles? Is that where the majority of the Western fanbase is at?
  16. You might consider giving AI-generated art a try. https://lunarmimi.net/mimi-art-tips/6-mind-blowing-ai-art-generators-for-anime-artist/
  17. Wow, sounds like a mess. I can overlook some of these issues, but these are pretty critical: I assume the game-breaking crash at the end will be fixed. These issues probably won't be, as there's no way they weren't noticed during testing. Not sure how they thought the game could be released with critical features broken or missing. I wonder if the skip function breaking is related to edits made to text in one part of the script that weren't properly carried over to duplicate text in other parts (resulting in previously read text not being flagged as already read). That would suggest either that their scripting code is spaghetti, or the edits were sloppy. I've seen this happen with old saves in serially updated Patreon games, where script and save versions don't match.
  18. I'm playing Yakuza 0 (on Steam) and there's literally a side-story where a ring of high school girls (led by a bully) are selling their used underwear. There's another side-story where an elementary schooler solicits the adult main character to be his girlfriend, and says she'll "give him her precious thing". And yet another where some punks try to sell a mother and her elementary-school child into prostitution. So, uh, yeah. Keep in mind: Sony NA is giving some of these titles a pass, while Valve isn't. So yes, Valve being stricter than Sony--while simultaneously carrying actual porn--is headline-worthy.
  19. I've been getting mixed messages about Shiravune as a English localizer/publisher. On one hand, many of their early releases were professional quality. On the other hand, I've heard that quality in their later releases has been inconsistent--they seem to have partnered with many different contractors of varying competence to increase throughput and reduce costs. Then there's the censorship: they've done some very weird things like censoring a shota protagonist out of the HCG.
  20. I tried Sugoi Translator (Offline) recently and was impressed by the strides made recently in machine-learning enhanced JP -> ENG translation. My friend and I independently checked the translation accuracy of the latest offline translation model (3.3) in two different titles. We agreed that accuracy has reached the point that those with some JP knowledge can use it as an alternative to dictionary lookups or for speed reading. Subject/object confusion is still a major problem and therefore we aren't quite at the point where it's safe/recommended for those with no JP knowledge. I ran some comparisons with DeepL and found Sugoi Translator (3.3 model) to be comparable to DeepL for VNs. Sugoi Translator has several significant advantages over DeepL: Can be run on the host computer with low translation latency (<1s) when properly configured. Optimized for VNs, and can recognize and translate odd speech patterns used by anime characters. Freeware with no rate limits. No online connection required. Doesn't spontaneously break all the time because you were blacklisted or the server API changed. The tool however is non-trivial to properly configure and will require some persistence and resourcefulness to fully optimize (e.g., update the model, use CUDA, and seamlessly incorporate into a workflow). Documentation is scattered and updates are frequent and not automatic, and therefore many users are probably using a non-optimized version. I'm sufficiently satisfied with Sugoi Translator that it's replaced ATLAS in my workflow. Until now I've found statistical machine translation alternatives either too unreliable (Google Translate) or too unwieldy (DeepL), but Sugoi finally convinced me that real progress has been made.
  21. The accuracy rate for dictionary words appears to be >90% [1]. What aspects of text hooking are useful to you likely stems from your own language knowledge. Clephas learned spoken Japanese first, so his main obstacle with reading VNs initially was linking written words to his extensive verbal vocabulary. That would likely be the main challenge for natives as well. For most of us, the dictionary aspect is far more useful than furigana because pronunciation is not necessary to extract meaning. In addition, claiming that dictionary lookup is more "accurate" than furigana parsing is not necessarily meaningful (words having meaning, right?). How would we assess accuracy? We'd look up the word and see if the right definition is somewhere in the list. Seems straightforward, but try this for 掛かる. This has exactly one reading, but 14 different definitions in EDICT. Just because the right definition is somewhere in there doesn't mean Beginning Language Learner A will be able to pick out the right one, or in any sort of expeditious manner. I can throw an unindexed 2000-page dictionary at you and say it's "100% accurate" because it contains every known definition for every word--but is that meaningful if you can't find what you're looking for? At least with furigana readings, the software attempts to pick out the right one rather than shove the entire burden on the language learner.
×
×
  • Create New...