- This topic has 58 replies, 14 voices, and was last updated 2 years, 7 months ago by
.
-
Posts
-
The natives are restless on the forums over Daz sticking ads on all their pages. See "Gallery Ads/Popups" thread. In its usual fashion there was no prior warning or discussion with its community. And not just ads but 3rd party, obtrusive click-bait ads in the gallery and everywhere else. Lots of members have now moved to ad-blockers or ad-free browsers to block them. Didn't Daz anticipate this and the blow back?
The forum mods didn't know it was happening and respond with "just have to live with it". The way this was done just continues the long list of arbitrary decisions that Daz has made and shows the level of disrespect it has for its members. It's almost like they're deliberately trying to kill the site.I feel that this 'Tafi' thing (their new owners?) has changed their direction. They announced Studio 5 then Tafi came along and everything changed. We got G9, NFT's and there will never be a Studio 5. It's going to stumble along until most of their artists are gone and they're losing money and they'll try to sell it. AI isn't gonna help them either.
It's a perfect example of enshittification.I know what that means IK and I did it 12 months ago. After 20 years I have little interest in DS these days.
The fact their own mods didn't know and were left to pick up the pieces shows it's not only us who are being shafted. Imagine working in a corporate environment like that.When (not 'if') Daz goes under, we're gonna be the keepers of the Realm. This will be the de facto preserve of the content and the many versions of Studio.
What an assume responsibility.Amusing to think of it that way @Ethiopia but very true. We bear a heavy responsibility 😁.
Between Daz hard to install content, unusable morphs when you have installed too many, unoptimized and archaic program, expensive bundles, requirements for a powerful computer, and overall subpar results (unless you are making something simple or using tons of Photoshop)... I have to admit that Daz doesn't have any fighting chance against AI imagery.
I have been at AI for the last 13 months, and besides requiring a good PC, everything is totally free (if you use Stable Diffusion), and you get incredibly fast updates (it's almost hard to keep up, in the last 10 days alone they announced 2 major AI image generation releases and a text to video one that looks insane, and honestly I cannot imagine how things are going to develop in the next 2 years...probably faster than I can imagine).
The time it takes for me to get "an acceptable image" with AI would be similar to using DS, ie, a few hours if I want a nice one. But the two processes are so different it's very difficult to compare the two methods other than to say neither is easy.
Personally I grew tired of the DS-look of the characters and renderers, being more into realism and fixability which AI does a much better at. With SD I can't make the scenes I used to make with DS but I can make images that DS can't pull off with respect to character/clothing realism, lighting, render time. While DS image generation is pretty predictable, pushing that Generate button on SD is like pulling the handle on a poker machine... no fk'n idea what will turn up. As a reformed gambler and ex-fisherman maybe that doubt and uncertainty is an added attraction for me.... old habits.
Rank:
Ok... Got some questions for you guys that have moved on-ish from DS.
I noticed your comment Frank about A1's unpredictability... I think that is my problem with it (though I have done little more than dabble with AI).
Lets say you were working on a game/VN, and needed the predictability... say, the same character over and over again in different scenes... Can AI do that? You can't have your main character change, even slightly from 'frame to frame' so to speak.
Also, it seems like AI is great at making a portrait, a 3/4 view of the character right in front of a camera, but if you need 3 characters building a house of cards, with the camera 10ft above and behind... AI not so much. Especially if you need the same three characters over and over.
There are ways to maintain a consistent character and clothing but I'll tag those questions for @Legolas18 who knows a lot more than me about those things. Multiple characters in one scene is a bigger problem and the bane of my existence.
I'm sure I'd read about folks making a basic scene in Studio, taking a screenshot then using this in SD to do Image2image.
@redflag While there are ways to get consistent characters with consistent faces and clothing, it's still impossible to have absolute control over an image to generate one image with 3 specific characters, or something very specific (which can always be trained in a LORA).
Things are changing very, very fast, though. SD3 has just been announced which is light years ahead in terms of prompt understanding, but it's still in early beta.
What I would do instead is generate characters with no background (and to be cut out as PNGs) with different poses/facial expressions and then superimpose them on a static background, as is typically done in VN 🙂
Iam still on the fence, cause AIs aren't there yet, at least not good enough for my liking. But the announcement of Sora and SD3 tho, things are changing faster than my expectations. We might have startrek's tier simulation(visually, atleast) in less than 5-10 years. Daz is going to be left behind further and further as we go. It's really sad.
@Ethiopia VN = Virtual Novel 😉
@Frank22 Several hours for a decent image? Hmm, if you like realism the best model for SD 1.5 would be the Epic Realism model (now also for SDXL), which would definitely generate better quality images on average.Definitely the graphics card has an important role in this, and if you have a "less powerful" one, I definitely recommend upgrading to Forge webUI, which is particularly geared to generate images faster with those cards and uses less VRAM too.
@Conkec1993 It really depends on what you want to create. For me, it's mostly Marvel or some fantasy girl in some Anime/graphic novel/ etc style, and AI is definitely there, although I do take the time to fix some minor stuff in photoshop (color balance, light balance, and a few "weird" stuff that happens here and there).
As for SD3 I am excited too, but new generations of models (as with the previous iterations like SD1.5 and SDXL) typically take 6 months of extra work of fine-tuning from the community to make it truly amazing.
- You must be logged in to reply to this topic.
