In relation to standards, simplification and auto LOD, unique to Daz...
JD_Mortal
Posts: 758
I would like an actual "Eye", which is created from a "Lathe format", so it binds to the "Eye-lids".
At the moment, the "Eye" (Eyeball), is just a special object, attached to a standard rig, a component of the head, not the eye-rig.
What I am suggesting, is that the eye, with all of it's standard components, become a "formulated component", to replace any "standard object" that normally represents the eye-ball and eye.
This "Formula", containing 95% of all desired potential variation ability, as it "crafts" the eye to desired scene-LOD and for geometric simplification. Naturally, since it is a 100% lathe-compatible item, a lathe-formula would be appropriate. This includes the clear cover of paper-thin tears and cornea-bulge, and a potential Lens and Retina, if desired.
The components being, "The yellows" (The discolored back of the eye-body), "The whites" (The semi-reflective fronts of the eye-body.), "The iris-transition" (The thin jointing area where the thick clear iris-muscles bind), The Pupil (Just the hole shape, consideration for non-circular holes here, lathed, but a modification to the output.), The Lens (if desired, for cataracts or glow, or red-eye reflections), The cornea (Formed to bulge-taste and diameter), The Retina (For those of us who can find use for the inner images, or reflections, or whatever.)
The standard eye being just the white/iris/pupil/bulge. (Bulge limited to LOD settings.)
With an "Eye-lid" target to the eye, the program can determine which "surfaces" actually need to be drawn, and the eye-lids can have the contact-edge, actually remain in contact by essentially being welded to the surface. (Not actually welded, as that isn't a possibility, but relocated to the intersect, which always points to the center of the eye, so it always collides with a surface.
The point being that in edit mode, it can be as low as 6-faces lathe-contour, with points matching the horizon and the flats being matched to the top and bottom lids... Or as high as an 80-sided lathe-contour in rendering, without any sub-division surfaces, or 80 with subdivision, if that can be done faster. (Obviously at the discretion of the program, based on the output of the "eye's size", in relation to the screen-render. It is a fast and easy formula to calculate, because it is essentially just a sphere.
With one model, the savings will be marginal... With dozens, the savings are rather great... With a hundred, the savings are more than welcomed.
The savings are the removal of the "Known" unseen faces, they are just not "formulated for output", thus, never exist. Along with the LOD, lathe-division. File-wise, this allows us to use a true HD EYE image that has the HD where it needs to be, in the IRIS, as the square image contains the ideal details per component, when unwrapped as a cylinder and wrapped onto the eye as a cylinder-over-a-sphere. No UV mapping is actually needed, since it is 360-deg from one edge on the 0 location. The mapping is output formulated too, on only the existing surfaces that needed it.
For the non-human eyes, the iris, and possibly the cornea should have an option to "pinch" the shape, which retains the circular bulge shape, but distorts the hole on the vertical or horizon. Devoid of the Iris, the "lens" can act as an Iris, for "toon eyes". Both the retina and lens would be a direct single flat surface, but built from the lathe, so it retains the contouring circular edge.
The "Lathe patterns", would be user-editable... for proportions of the Yellow, White, Iris, Pupil, Bulge... etc... Just a simple ratio to remember, for every eye that is "not default". Less data than trying to remember 5,000 points and UV mappings of something that a formula can output within a second. Per model... Per LOD... Even the "default eyes" are retained in every save. (Redundant, it hasn't changed.)
For those who have "special eyes", beyond the requirements of this "smart component"... They just turn it off, and play with triangles again.
Next we'll talk about the ears... Organic formulas fit well into your morphs. Might as well take advantage of them, and what they naturally offer.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
What makes you think any saving from slightly simpler geoemtry would not be more than wiped out by generating and managing the procedural? What problem is this additional complexity meant to be addressing?
Storage space, mapping standards, universal components. (Specific only to the eye, as that one shape is universal for almost every biological and fantasy "animal".) Since this is reverse compatible, and forward compatible and adaptable, that is one less thing to "design" and "adjust" and "learn". Once it is learned, it is universal to everything. Importing, exporting, display, LOD, pre-processing, optimization.
10,000 models have 20,000 eyes, with 3 LOD shapes, leaving 60,000 unique-saved non-unique items. And even when 90% of them are the exact same shapes and mappings and structure, there are 60,000 individual objects saved as "Redundant information", which all require "Manual upgrades and downgrades per each of the 60,000 items", for LOD, as models become more advanced or situations call for less details.
Why not make life more simple, instead of more complex and demanding of efforts that computers were designed to save us time for... Like reinventing the wheel/eye.
Ears are more of a trick, as genetics per "animal", result in variations that can make components too unique. However, the human ear, like the human nose, has specific attributes that do NOT change, and exist in every human. With rare exceptions to those "special" individuals who are the rarest of breeds. Just as we have the same, for the most part, hands and feet. (Sort of the point of having core-models to morph from. Much redundant data is stripped from morphs, but they all depend on the core-model, which keeps changing... V1, V2, V3, V4, V5-Gen1, V6-Gen2, V7-Gen3... And even those could have, potentially, been a seamless upgrade, but the choice was made to essentially make each one from scratch, or make them incompatible, for whatever reason. They could have been made 100% compatible, if it was orchestrated that way. Obviously, it wasn't even a consideration.)
Beyond the eyes, Nails could be this way, if desired. When the time comes that DAZ includes nails as a unique object. With a simple universal code that handles every form of nail from human to dog to dragon to horse-hoofs. (the hair-tool does a similar thing. However, it is dealing with millions of strands, not two eyes or ten nails, per model.
The Daz library has thousands and thousands of items, IMVU-Chat has millions and millions, using a similar form of "Deriving" which leads to HUGE savings. (However, they do not use the eye-trick I just mentioned, because no-one that I know-of does.. because I just thought it up. :P... It's shoes for your feet, that look like feet...
For me, it would save me the time of having to position the eye-lids, having to ensure they blink correctly when closed, ensuring they touch the eyeball without the bulge poking through, and save me time from attempting to re-map an eye to some deformed flat orbit UV that doesn't retain detail where I need it, and save time from having to reconfigure every polygon to compensate for larger Iris's, or non-human iris's, and having to setup special "watery-eye layers", and "bulge layers". For 1000 unique individuals, as each item gets saved uniquely, and morphs don't compensate or adjust for this stuff. When they could.
Also, simple realism.
Fake eyes on real bodies, like fake hair, is detracting from the quality of the images and the products.
I understand that Hair is a tough item to tackle, but eyes, which are made overly complex, for no apparent reason, are just adding difficulty without providing much actual realism to justify the expended hours needed to make adjustments or craft whole new eyes from scratch.
With a lathe-item, you only save the profile... In this case, the profile is nothing more than ratios per component, as they all produce an Eyeball, which is a specific universal shape. Though the unique components change location, as the bulge-size, bulge-position on the surface, iris diamater, etc... which are all part of those proportions, along the spherical lathed-component. (The eye is actually not perfectly round, but has a slight oval shape to it too.)
In addition, with the eye-lid points being part of the "Eye component", the external "eye-lids" know where to weld/connect. Closed is always closed, not overlapping. 50% open is always 50% of the eye-balls "max-open" which should be about half the orbit, or the eye would pop-out... which may be desired, and that could be possible realistically, at 100% open... but the eye-lids OBJ which are NOT part of the eye, would KNOW where they have to move to, to facilitate this "Open" state, and never "lift-off" or "slide inside" of the eyeball, as the same code that tells the eyeball where to draw the surface, is also positioning the eye-lid attachement points. (The attachment points could be controlling 100 points along the eyelid-object, but only six points are needed for the rig. The two corners and two positions along each of the lids, top and bottom. Obvioulsy, user editable, but they always remain at the surface of the eyeball, that radius from the center, plus any deformity from the "bulge". Which is faster to calculate than trying to do collision per 1000 points along the eye-lids, with the "smooth-node". Which often guesses wrong when items are that-close or limited in polygons. Since three matches the minimum needed for an open eye-lid, that is perfect for reverse and forward compatability. Later, more can be added between for more control, without altering the original points. Still reverse compatable and now forward compatable for more precision.)
This is a trick from animatronic eyes, where the Rig for the eye is attached to the eyeball, with the eye freely moving within the rig, and the eyelid-controller controlling the "skin" of the model by simply gluing it to the rig. (Welding/connecting).
But alien eyes, human, dolphin, toad, deer, anime... all work the same. Just the shapes may be proportionally different, as stated above. A toad would have nearly no white, and a huge bulge. A deer similar... Yet, an alien may only have a tiny iris or pupil that is rice-shaped like a snake-eye. (They don't blink... snakes, not aliens... everyone knows aliens blink... They made us!)
This code is no more complex than the code that handles drawing primitive spheres. However, this is a primitave we ALL can actually use, in its primitave form. Except the guy making robots without eyes, and furniture, and clothing, and landscapes... But they aren't using primitave spheres anyways... no-one is... (Ok, some guy is, for testing sphere ray-tracing... :P)
How are you getting tens of thousands of figures? Each figure in DS (or Poser, for that matter) exists as only one geometry file (and in the case of DS only one copy of the rigging and morphs), the character files contain only settings (in most cases - there are situations where customised geoemtry or morphs may have a local, non-asset, copy).
There is not just one model for everything. There are tons of individual models, in Daz, on my computer, all over. All with eyes. My eye morph, contained a whole eye, though it didn't actually modify the eye, just because it moved.
However, that wasn't the point. Every eye variation, on top of the eyes within the individual unique models, are also whole models. Models that later get unused because they get replaced by higher LOD models or new designs, or whatever...
Those, like the model variations, could also be just another universal asset, like V3 itself, and V4 itself, and V5 itself, all have unique eyes that could be unified assets, by the way.
At the moment, daz has, at-least 7x 2 = 14 unique models. Way more actually. v1m, v1f, v1u, v2m, v2f, v2u, v3m, v3f, v3u, v4m, v4f, v4u, v4.1m, v4.1f, v4.1u, v4.2m, v4.2f, v4.2u, g1u, g1m, g1f, g2m, g2f, g3f (Plus all the conversions, which are unique objects, morphed to shape, then "fit", which modifies the eyes to the point of being whole new objects.)
(M=male, F=Female, U=Unimodel) They are not just morphs of each-other. Those are whole individual models, not depending on the other for use. (Though the 4.2 is the new 4, the old 4 still exists, as does 4.1, because not everything "fits" the conversions and upgrades.)
Plus, it's another "blink morph" not cluttering the variations, and supplemental "corrections", because it wouldn't need corrections. Closed is closed, and open is open. That never changes from one animal to the next, or alien, except for snakes that don't blink and aliens with four-eye-lids.
Standards and simplification.
Where are the corners of the models eyes... Great... Where is the top lid... Great... Where is the bottom lid... Great... Adjust it a little... Poof, instant eyeball that just works as expected.
The human mouth could be done the same way, but that will require unique variations that are beyond "formulas", other than the fact that a mouth has an orifice with a representative set of "lips", that have a specific universal "closed" position, and a variable "open" limit/shape/form. What's inside is what changes, but that is not rig related. The only universal portion of a mouth would be the rig. Which the existing "rig", is more of a generic hack, as opposed to an actual universal rig. Floating points in space that have no actual linked relations or realistic defined points and constraints or limits. Better than what most places offer, but nothing related to standards.
Only the bold entries are ral, with unique mesh. Though you also missed Stephanie and Aiko (2), Stephanie 3, Aiko 3, Hiro 3, Freak 3, David 3, V2P4, V3 V2, A3 SP, and V4 V3 plus sundry RR versions.