Hello Friend,

If this is your first visit to SoSuave, I would advise you to START HERE.

It will be the most efficient use of your time.

And you will learn everything you need to know to become a huge success with women.

Thank you for visiting and have a great day!

3D Modeling / Rendering / Compositing / Animation

Deep Dish

Master Don Juan
Joined
Nov 25, 2002
Messages
2,154
Reaction score
143
This post has been in my head for a very long time.

There are four core skills to being a 3D artist: photography, drawing, painting, and coding/mathematics.

You learn so much from photography, from composition to the science of light. Understanding camera settings and lenses is necessary when rendering and compositing.

It’s not absolutely necessary to learn how to draw to be a 3D modeler, because you can model from blueprints and reference photos, but it’s helpful. Your drawing doesn’t need to be polished production art (a lot of “concept art” released by game studios is actually promo art rather than the early concepts), but you do need to be able to convey ideas. Drawing and modeling is a positive feedback loop where drawing improves your modeling and your modeling improves your drawing. Sketches are still faster than modeling (your mileage may vary). It’s so easy to get bogged down by technical minutiae in 3D. Speed is crucial because ideas vanish. After you have made a pass of your model and wonder where to go next with it, or how it can be better, draw over a screenshot with more ideas. When you’re happy with your model, you can do a paintover in a painting.

The fourth core skill of coding/mathematics is separate from the rest, because very few artists are both artistically and technically-minded. If you are one of those rare artists, you gotta try SideFX Houdini.

CGI is the most complex artistic medium which is both the challenge and appeal of it. It’s a good idea to stick to one discipline as your primary focus – concept art, modeling, texturing, rigging, animation, or compositing, and so forth – and achieve competent mastery before moving your focus onto another. It’s so easy to lose focus and you don’t want to be a jack of all trades but a master of none. There are artists who are good at everything, but it takes many years of hardcore practice.

When learning modeling, start with small and simple projects. Gradually work your way up to more complexity with every new project. You don’t want to bite off more than you can chew. Many projects die to grave when you hit a hard spot where you don’t know how to model something or run out of energy because the scope of the project was too ambitious for your level of experience. 3D is notoriously hard but the hardest part is finishing a project.

Making a drawing, painting, or model has three stages: primary forms, secondary forms, and tertiary fine details. Big, medium, small.

The quality of a project is directly correlated with how much passion and time you put into it. There’s a maxim that the last 20% of a project takes 80% of the time. It’s easy to fall into the trap of thinking that you’re almost done when you really have hundreds of hours left to go. How long a project takes depends on the level of detail.

It’s best practice to always model to real world scale, when possible. Even if a model is from your imagination which gives you creative liberty, it’s best to figure out its size. In 3D, scaling is not a simple action of pushing a button, because problems can arise when the scale of a model is changed. Rigging for animation and wall thickness for 3D printing will break, and game engines are slowed down by calculations. That said, real world scale is not always possible.

Sculpting someone’s likeness is extremely hard. You can be very successful at sculpting a human but just not who you were going for. You need 100% perfection. If you’re off by even one millimeter, it ruins the illusion. You can sculpt a lion and get away with mistakes, but we are experts at recognizing humans. To make things worse, your reference photos were taken in different poses at different ages with different focal lengths and the need for multiple photos makes it very difficult to sculpt a perfectly anatomically correct model, other than taking the photos yourself for photogammetry. Your goal is to capture the essence of an anatomical average.

Unless you do your own photo shoot, it’s recommended that you sculpt A-list celebrities because they have a lot of photos. You should also watch videos because there are things you can miss from just photos. If you’re doing an actor, it’s a good idea to take references from one movie or television season so they are the same age.

You need as many reference photos as you can get, but you need at least three photos: front view, side profile, and three quarters.

Link: “21 tips on capturing a human likeness”

Study anatomy. You’re not a doctor and so you don’t need to know every bone and muscle, but knowing the names makes it easier to remember the anatomy and communicate with other artists.

Likeness doesn’t really come in until you’ve done the hair.

There are a few ways to check a model. The first way is to look at the silhouette, which will tell you if the primary forms are working. The second way is to flip your model upside down, it resets your eyes by tricking your brain into thinking it’s looking at a new pattern. It’s also good to step away from a project for a days to come back with fresh eyes. When you’re staring at something for a long time, you become blinded to mistakes. The third way is to overlay a photo, but be mindful that because of differences in focal lengths between the photo and your 3D modeling package that you will never have a 1-to-1 match, unless you can pull the focal length of the photo from EXIF data or adjust the 3D camera. Photos with longer focal lengths have less distortion.

MODELING

There is a modeling trick to split off polygons from a mesh into its own object. If you’re modeling a building, break off the walls before doing the windows. This allows you to keep the geometry simple without needing to manage edge loops for proper topology. However, there are times you will need one watertight mesh. If you place lights into the rooms of a building, you will experience light leaks if the walls are not connected, and 3D printing needs to be watertight. It’s often easier to model something as separate parts and then merge it together later if necessary rather than modeling it as one solid piece from the beginning.

Use instances for duplicating objects. Instances are not real geometry and are pointers. The two advantages are that it saves a lot of memory and the duplicates are instantly updated when the original geometry is changed. You can have 15,000 instances in a scene where 15,000 objects would choke any 3d modeling package (even if each object was only one polygon).

Use bump maps, displacement maps, and normal maps. This allows you to fake geometry and keep the polygon count low.

You can do some modeling with 2D illustration programs and export as SVG (scalable vector graphics).

For sculpting, you can use dynamic tessellation (Sculptris Pro in ZBrush and Dyntopo in Blender) to block out your shapes. Dynamic tessellation allows you to freely sculpt without worrying about stretching polygons and is the closest experience to sculpting with clay. However, it’s only good for the early stages of a sculpt and you will need to retopologize it before texturing or rigging. It doesn’t work with subdivision levels and doesn’t work with high polygon counts.

Make a primitive in a 3D modeling package before bringing it into ZBrush, so you when you’re done with ZBrush and bring your model back into the scene, it will be the correct scale and position in 3D space. If you start in ZBrush, you will have scaling problems when bringing it into a scene.

DON’T USE A SMART PHONE CAMERA

Smart phones are good for reference photos, but the focal length of the camera lens on a smart phone is different than the focal length on a real camera for the same photo. This creates a problem because for compositing you need to match up the focal length of the camera lens in the 3D modeling package with the focal length of the photo. For photogammetry, you need the biggest megapixel resolution that you can get your hands on (contrary to general photography advice that megapixels don’t matter) and smart phone cameras are usually (but not always) low resolution. In theory, a Hasselblad camera with 400 megapixels is the best possible camera for photogammetry, but how many people can afford a $48,000 camera?

DO THINGS THE HARD WAY

There are tools which make your life as an artist easier. Things like premade assets (from other artists), DAZ Studio or MakeHuman for character creation, and Marvelous Designer for cloth simulation. It saves you a lot of time, in a production you need to use every trick in the book, but if you are a beginner, it’s a mistake.

Learn how to do things the hard way, so you know how to do it. Even though premade assets save you a lot of time, and in the professional world you’re not expected to make everything yourself, you need the learning experience. You can do more when you know how to do it yourself. When you understand drapery, for example, you can go beyond reference photos. Learn the foundational skills of art and you will be a free spirit, but skip ahead to the tools and you will always be a slave of your tools.

Premade assets (from other artists) are also expensive, so you can save a lot of money by modeling things yourself. One great advantage of 3D modeling is that you only have to model something once. Over the years, you build up a whole world of things that you can reuse or repurpose in your projects. There is a whole genre called kitbashing.
 

Deep Dish

Master Don Juan
Joined
Nov 25, 2002
Messages
2,154
Reaction score
143
PHOTOBASHING

People often assume that using photos is easy and some think it’s “cheating.” There’s no “cheating” in art (other than plagiarism) and photobashing is standard practice in commercial art to quickly develop concept art in hours rather than a month for a painting. It’s hard – not as hard as painting, but still – because you can’t use just any photo that you want and you still need to do a good amount of painting. You do need painting skills as there are things which are not easily done with photos or 3D.

When using photos, you need to be mindful of matching the lighting, perspective, and color temperature. For color temperatures, Photoshop has a feature called Match Color.

Removing one or two people from a photo is easy peasy, removing a whole crowd is tricky because so much of the background environment is blocked from view. There is the photography trick of using a tripod to snap multiple photos to stitch together areas of no people, but traveling on-location is expensive.

For removing objects from photos, there is content-aware fill and the clone stamp tool, but often you need to get dirty with painting it out by hand. It’s important to do things on separate layers so you remain non-destructive. After using the clone stamp tool, you can paint out details to make it less obvious that things were duplicated.

The best photobash artists do so much painting that you can’t even tell that photos were used.



ENVIRONMENT MODELING

There are two approaches to modeling an environment.

If the model is going to be a still image render, model everything from one camera angle. Figure out the camera focal length, aspect ratio, camera perspective, and then be sure to lock the camera. It’s important to solve for the camera early on, because you don’t want to spend a lot of time modeling details which will be far from the camera. This is one place where drawing skills are helpful because a drawing automatically solves for the camera and the frame of the drawing for the aspect ratio.

There is a cool compositing trick called camera projection (or projection mapping), which is when you project images onto 3D geometry to create a parallax effect to give a 2.5D illusion of 3D depth animation of small camera movements in a photo or matte painting. The first movie with camera projection was Steven Spielberg’s Hook in 1991. Even if you have 3D models, camera projection can save time for a small animation because you only need to render one frame.


The other approach is to model an environment from multiple camera angles or all 360 degrees. It’s very time consuming. I highly recommend using the desktop version of Google Earth Pro because you can take 3D measurements in cities which have been 3D scanned. 3D View is good for primary shapes and Street View for secondary shapes and tertiary fine details. I also recommend enabling GPS location tagging on your camera so that when you find a cool location in the real world you can easily find the exact spot again in 3D.

When modeling an environment, and really when modeling anything big, it’s crucially important that you import a model of a human (or something else of a known size) into the scene to establish the scale.

BREAKUP SURFACES FOR VISUAL INTEREST

You want a ratio of about 70%-30% between high frequency detail and low frequency detail. You don’t want 100% high detail because your eyes need areas of relaxation. You don’t want equal balance because it’s boring. Just as with the Rule of Thirds, 70%-30% is an approximation of the Fibonacci Golden Ratio.

AVOID TANGENTS

Tangents are sneaky little fückers. There’s so much to think about during composition that tangents are easy to miss. Tangents happen when the edges of objects touch each other or touch the frame of the image, or are grouped near each other. Tangents are distracting and create visual confusion.

REBATMENT OF THE RECTANGLE

Rebatment is similar to the Rule of Thirds. Within every rectangle is a perfect square.

Speaking the Rule of Thirds, some people get hung up on the word “rule,” even so far as making a Youtube video calling it a myth. No, it’s not a “rule,” it’s a guiding principle. If you find a composition trick like symmetry works better for an image, go ahead and use that. While not a “rule,” Rule of Thirds is a great starting point.

DEPTH IN COMPOSITION

You probably know about foreground and background. Well, there’s also middleground.

In a composition, add something in the foreground which is close to the camera but off to the side, like a tree or car, to create more sense of depth. When you have all three, the middleground has saliency.

Another thing which adds depth is lighting. Your eyes are actually more sensitive to light than color. By having contrast of lighting between light and dark, you can extend the sense of depth.

Repeating elements and atmospheric particles also add depth.


HARDWARE

CPU cores are a balancing act. You want as many cores as you can afford, but more cores mean higher electricity bills every month. (Workstations are just as effective at warming up a room as a space heater (link)) Most tasks are single threaded, so most of the time you don’t need multiple cores. Multiple cores come into play when rendering, when having multiple applications open at once, photogammetry, and probably some other tasks.

The GPU is the most important piece of hardware. You need to throw as much money into it as you can afford. For NVIDIA cards, there is Quadro and GeForce. Even though Geforce is made for gaming and Quadro for workstations, Quadro is slower than GeForce because Quadro is optimized for precision rather than speed. If you’re a designer with CAD work, you really need precision. But for game development or rendering, it makes more sense to go for GeForce even on a workstation. You can get faster performance for your money from AMD but AMD GPUs are not suitable for 3D work. I don’t know the details except to say AMD GPUs have poor quality drivers. NVIDIA, which is the spiritual successor of Silicon Graphics, caters more to professional 3D artists.

You need as much RAM as you can afford. There’s no such thing as too much memory. Basic modeling is fine with 8GB, but more RAM allows you to model more complex scenes and photogammetry can possibly consume upwards of more than 700 GB (link.) VFX artists often have 256 GB. I personally have 32 GB of RAM.

RENDERING

The most common words you will come across: diffuse is color, specularity is reflection, roughness is the sharpness of specularity, ambient occlusion is contact shadows, global illumination is indirect bouncing of light, subsurface scattering is translucency, and volumetric lighting is fog or haze.

Bigger lights have softer shadows.

For perfectly accurate recreation of real world lighting, you can use IES lighting profiles which are provided by the manufacturer of your light bulb.

In Cinema 4D, you have the rendering options of Irradiance Cache and Monte Carlo for global illumination. Irradiance Cache is faster because it does a pre-pass to map out areas, but it’s a pain in the ass. In the words of Arnold documentation, “What is wrong with algorithms like photon mapping or final gather? Such algorithms attempt to cache data that can be re-sampled later, to speed up rendering. However, in doing so, they use up large amounts of memory, introduce intermediate steps that break interactivity, and introduce bias into the sampling that causes visual artifacts. They also require artists to understand the details of how these algorithms work to correctly choose various control settings to get any speed up at all without ruining the render. Worse than that, these settings are almost always affected by other things in the scene, so it’s often possible to accidentally use settings for the cache creation/use that make things worse, not better, or that work fine in one situation but are terrible in another, seemingly similar, situation. In short, they are not predictable, other than for very experienced users, and require artists to learn way too much about the algorithms to gain any benefit. We believe that your time is more valuable than your computer’s time; why spend an extra 30 minutes working with photon mapping or final gather settings, even if it saves 30 minutes render time (and more often than not it doesn’t).”

It’s best to render image sequences of individual frames rather than directly to a video format, because it’s higher quality. Video is only your final deliverable. The best image format is EXR because it allows you to save all render passes in a single file, although the files get huge.
 

Deep Dish

Master Don Juan
Joined
Nov 25, 2002
Messages
2,154
Reaction score
143
Don’t use a laptop for rendering. Most laptops are just not designed for very heavy hot workloads for long periods of time. While rendering is possible with mobile workstations, it has short battery life because of intense power consumption and you can always get more machine for your money with a desktop.

After the render is done, you will need to use photobashing/photomanipulation techniques to complete your render. At minimum, the render will need color correction. Adding photos adds nice touches of detail and painting smoke is easier than 3D simulation. If modeling something large like a building, you will almost assuredly need to do photo editing, because finding the perfect photo for a background is close to impossible, unless you are using a game engine for visualization.


RAY TRACING VS. PATH TRACING

Ray tracing traces rays of light from the camera to the light source. This backwards approach is more efficient than tracing rays from a light source to the camera because many rays would be wasted by going somewhere the camera can’t see. With ray tracing, when a ray hits an object, rays are sent to every light source. Path tracing is a kind of ray tracing, but when a ray hits an object, it simply bounces the ray in a random direction.

Arnold (which is what I’m currently using) is a Monte Carlo path tracer.

UNBIASED VS. BIASED RENDERS

Unbiased render engines use brute force and treat every ray the same to achieve a physically accurate result (on average). These include Arnold, Keyshot, Octane, Mantra, and Cycles. Render engines like Mantra are optimized for large scenes where Cycles is optimized for small scenes. Mantra is painfully slow but that’s because Houdini can handle MASSIVE amounts of data. Biased render engines use tricks like ambient occlusion to achieve faster results and can look just as physically accurate (on average). These include V-Ray and Redshift.

RETOPOLOGY

Retopology is the re-surfacing of a high polygon model into a low polygon version. ZBrush makes amazing models but using the high polygon model in another 3D modeling package is not a good idea. There are tools like Zremesher in ZBrush which do auto-retopology for you, but you still need to learn how to do it manually. Zremesher is great for still image renders or objects in an environment which will never move. Sadly, however, there is no such thing as auto-retopology for animation, because it’s not good for rigging. Even if you spend a long time prepping a model to guide a Zremesh, you will run into more problems when rigging, nothing beats doing it by hand.

For animation, topology needs to be quads (four-sided polygons). All polygons are converted into triangles at render time, but triangles have unpredictable effects. It’s often impossible to avoid triangles completely, so try to keep them to a minimum and away from areas of animated deformation (with an exception of collapsible knee joints). Topology needs to follow the edge flow of muscles.

Link: “Face modeling”

Speaking of triangles, polygon counts in game engines are measured in triangles. If your model has 10,000 polygons in quads, you actually have 20,000 polygons as far as the game engine is concerned.

MOCAP

Sorry, but motion capture is not a shortcut to animation. You need to cleanup the mocap data, so you still need to be a skilled animator. It saves time, but not nearly as much as you would think.

GREEN SCREENS

Green screens should only be done if you are going to be dedicated to it, because it gets expensive. Cheap kits generally have horrible reviews (except Limostudio is decent) and, even still, you will probably need additional lighting equipment than comes with a kit. You may not be able to use lights from around your house because of color temperatures.

The history of chroma key started with blue screens (before chroma key were mattes and traveling mattes), and Disney experimented with yellow from sodium vapor, but blue switched to green when digital cameras were invented because digital sensors are more sensitive to green. Green screens are good for bright scenes and blue is better for low light.

Speaking of lighting equipment, C-stands (or “century stand”), which are very versatile and highly recommended, were first used in the very early years of film. Before studio lights were invented, stages were rotated throughout the day to stay aligned with the sun and C-stands were used to bounce the sunlight onto the stage. (Be careful with handling C-stands, it’s dangerous.)

Screens need to have no wrinkles, you need to be at least six feet away (because of light spill), and the lighting needs to be even to avoid hotspots. Even though the lighting may look perfectly even to your naked human eye, it’s not. You need to use a field monitor to make sure the lighting has no hotspots. When done right, chroma key is one click of a button in post-production. When something goes wrong, it’s better to re-shoot rather than the nightmare of trying to fix it in post.

LAYER VS. NODE-BASED COMPOSITING

You can use layer-based After Effects, it’s just as capable. However, I recommend considering nodes.

Layer-based compositing is better for simple composites, where nodes are better for complex composites. You going to need to become comfortable with nodes, because nodes are used throughout high-end 3D workflows, from procedural modeling and procedural texturing to VFX. With nodes, you’re able to experiment with different ideas without needing to press the undo button and easier to troubleshoot problems by tracing through the wiring of nodes instead of soloing elements as you would with layers.

Foundry Nuke is the best node-based compositor but it’s exorbitantly expensive. I recommend Blackmagic Fusion which is paired with Davinci Resolve (the best color grading) and much cheaper in the long run than an Adobe subscription of After Effects and Premiere Pro.

So there you go.
 

Ohso-Phresh

Senior Don Juan
Joined
Feb 22, 2020
Messages
368
Reaction score
279
This post has been in my head for a very long time.

There are four core skills to being a 3D artist: photography, drawing, painting, and coding/mathematics.

You learn so much from photography, from composition to the science of light. Understanding camera settings and lenses is necessary when rendering and compositing.

It’s not absolutely necessary to learn how to draw to be a 3D modeler, because you can model from blueprints and reference photos, but it’s helpful. Your drawing doesn’t need to be polished production art (a lot of “concept art” released by game studios is actually promo art rather than the early concepts), but you do need to be able to convey ideas. Drawing and modeling is a positive feedback loop where drawing improves your modeling and your modeling improves your drawing. Sketches are still faster than modeling (your mileage may vary). It’s so easy to get bogged down by technical minutiae in 3D. Speed is crucial because ideas vanish. After you have made a pass of your model and wonder where to go next with it, or how it can be better, draw over a screenshot with more ideas. When you’re happy with your model, you can do a paintover in a painting.

The fourth core skill of coding/mathematics is separate from the rest, because very few artists are both artistically and technically-minded. If you are one of those rare artists, you gotta try SideFX Houdini.

CGI is the most complex artistic medium which is both the challenge and appeal of it. It’s a good idea to stick to one discipline as your primary focus – concept art, modeling, texturing, rigging, animation, or compositing, and so forth – and achieve competent mastery before moving your focus onto another. It’s so easy to lose focus and you don’t want to be a jack of all trades but a master of none. There are artists who are good at everything, but it takes many years of hardcore practice.

When learning modeling, start with small and simple projects. Gradually work your way up to more complexity with every new project. You don’t want to bite off more than you can chew. Many projects die to grave when you hit a hard spot where you don’t know how to model something or run out of energy because the scope of the project was too ambitious for your level of experience. 3D is notoriously hard but the hardest part is finishing a project.

Making a drawing, painting, or model has three stages: primary forms, secondary forms, and tertiary fine details. Big, medium, small.

The quality of a project is directly correlated with how much passion and time you put into it. There’s a maxim that the last 20% of a project takes 80% of the time. It’s easy to fall into the trap of thinking that you’re almost done when you really have hundreds of hours left to go. How long a project takes depends on the level of detail.

It’s best practice to always model to real world scale, when possible. Even if a model is from your imagination which gives you creative liberty, it’s best to figure out its size. In 3D, scaling is not a simple action of pushing a button, because problems can arise when the scale of a model is changed. Rigging for animation and wall thickness for 3D printing will break, and game engines are slowed down by calculations. That said, real world scale is not always possible.

Sculpting someone’s likeness is extremely hard. You can be very successful at sculpting a human but just not who you were going for. You need 100% perfection. If you’re off by even one millimeter, it ruins the illusion. You can sculpt a lion and get away with mistakes, but we are experts at recognizing humans. To make things worse, your reference photos were taken in different poses at different ages with different focal lengths and the need for multiple photos makes it very difficult to sculpt a perfectly anatomically correct model, other than taking the photos yourself for photogammetry. Your goal is to capture the essence of an anatomical average.

Unless you do your own photo shoot, it’s recommended that you sculpt A-list celebrities because they have a lot of photos. You should also watch videos because there are things you can miss from just photos. If you’re doing an actor, it’s a good idea to take references from one movie or television season so they are the same age.

You need as many reference photos as you can get, but you need at least three photos: front view, side profile, and three quarters.

Link: “21 tips on capturing a human likeness”

Study anatomy. You’re not a doctor and so you don’t need to know every bone and muscle, but knowing the names makes it easier to remember the anatomy and communicate with other artists.

Likeness doesn’t really come in until you’ve done the hair.

There are a few ways to check a model. The first way is to look at the silhouette, which will tell you if the primary forms are working. The second way is to flip your model upside down, it resets your eyes by tricking your brain into thinking it’s looking at a new pattern. It’s also good to step away from a project for a days to come back with fresh eyes. When you’re staring at something for a long time, you become blinded to mistakes. The third way is to overlay a photo, but be mindful that because of differences in focal lengths between the photo and your 3D modeling package that you will never have a 1-to-1 match, unless you can pull the focal length of the photo from EXIF data or adjust the 3D camera. Photos with longer focal lengths have less distortion.

MODELING

There is a modeling trick to split off polygons from a mesh into its own object. If you’re modeling a building, break off the walls before doing the windows. This allows you to keep the geometry simple without needing to manage edge loops for proper topology. However, there are times you will need one watertight mesh. If you place lights into the rooms of a building, you will experience light leaks if the walls are not connected, and 3D printing needs to be watertight. It’s often easier to model something as separate parts and then merge it together later if necessary rather than modeling it as one solid piece from the beginning.

Use instances for duplicating objects. Instances are not real geometry and are pointers. The two advantages are that it saves a lot of memory and the duplicates are instantly updated when the original geometry is changed. You can have 15,000 instances in a scene where 15,000 objects would choke any 3d modeling package (even if each object was only one polygon).

Use bump maps, displacement maps, and normal maps. This allows you to fake geometry and keep the polygon count low.

You can do some modeling with 2D illustration programs and export as SVG (scalable vector graphics).

For sculpting, you can use dynamic tessellation (Sculptris Pro in ZBrush and Dyntopo in Blender) to block out your shapes. Dynamic tessellation allows you to freely sculpt without worrying about stretching polygons and is the closest experience to sculpting with clay. However, it’s only good for the early stages of a sculpt and you will need to retopologize it before texturing or rigging. It doesn’t work with subdivision levels and doesn’t work with high polygon counts.

Make a primitive in a 3D modeling package before bringing it into ZBrush, so you when you’re done with ZBrush and bring your model back into the scene, it will be the correct scale and position in 3D space. If you start in ZBrush, you will have scaling problems when bringing it into a scene.

DON’T USE A SMART PHONE CAMERA

Smart phones are good for reference photos, but the focal length of the camera lens on a smart phone is different than the focal length on a real camera for the same photo. This creates a problem because for compositing you need to match up the focal length of the camera lens in the 3D modeling package with the focal length of the photo. For photogammetry, you need the biggest megapixel resolution that you can get your hands on (contrary to general photography advice that megapixels don’t matter) and smart phone cameras are usually (but not always) low resolution. In theory, a Hasselblad camera with 400 megapixels is the best possible camera for photogammetry, but how many people can afford a $48,000 camera?

DO THINGS THE HARD WAY

There are tools which make your life as an artist easier. Things like premade assets (from other artists), DAZ Studio or MakeHuman for character creation, and Marvelous Designer for cloth simulation. It saves you a lot of time, in a production you need to use every trick in the book, but if you are a beginner, it’s a mistake.

Learn how to do things the hard way, so you know how to do it. Even though premade assets save you a lot of time, and in the professional world you’re not expected to make everything yourself, you need the learning experience. You can do more when you know how to do it yourself. When you understand drapery, for example, you can go beyond reference photos. Learn the foundational skills of art and you will be a free spirit, but skip ahead to the tools and you will always be a slave of your tools.

Premade assets (from other artists) are also expensive, so you can save a lot of money by modeling things yourself. One great advantage of 3D modeling is that you only have to model something once. Over the years, you build up a whole world of things that you can reuse or repurpose in your projects. There is a whole genre called kitbashing.
Dropping some domain knowledge, I like it.
 
Top