Managing Poser Scenes (14. Indirect Lighting [IDL] )

Indirect Lighting, aka IDL, is a computational intensive lighting strategy which can be considered the successor of IBL, Image Based Lighting. The use of it can be switched on/off in Render Settings. The basic principle of IDL is that loads of light rays travel around through the scene, hitting objects, and be re-radiated by those objects again usually with an adjusted color and intensity. This supports ambient lighting, indoor lighting from outdoor sources, radiating objects, radiosity (colors spilling over to neighbor objects) and proper mild shadow levels.

For a start, just a collection of notes and remarks:

* IDL is a successor of IBL, easier to use but far more computational and memory intensive. So, when IDL is not really working for you in a specific scene or on specific objects, consider re-introducing IBL as an alternative.
* Like IBL, IDL is working on Diffuse channels only, including Reflection, Refraction, Alternate_Diffuse. It is explicitely not working on Specular (and Alternate_Specular, and any specular material node wherever in the shader tree).
* IDL lights do not show in preview. As a result of this, and of the previous point, consider to use direct light in the scene with the Diffuse disabled (blackened out). And eventually, with Specularity blackened out too.
* IDL renders best (for final results) with Irradiance Caching ON, value at least 80 at most 90, and with IDL ON, Quality at least 80 at most 90. Lower values introduce noticeable splotches all-over the image, and overly dark shadows in self-shadowing areas. Higher values take a lot of time and resources while not adding noticeably to the result.
* It should be clear then that IDL does require raytracing to be active. This also introduces another mechanism to let the light rays die: when the limit for bounces is met, as set in Render settings, Poser cuts off any further handling of them. This will darken the ambient lighting, might introduce artifacts, and is explicitly meant for speeding up draft renders. Please set Bounces to the max when making your final render.

An interesting point is: I can launch the rendering from the Dimension3D menu too, and get access to additional settings. Indirect Light does have its own Bounces and Irradiance Cache, next to the generic ones for AO and reflection / refraction.

* IDL renders best when the scene is enclosed by a sky dome, walls of a room, or anything alike that traps the light rays and keeps them bouncing around.
* IBL comes with AO (Ambient Occlusion) to improve on shadowing from ambient, environmental lighting. Any other direct light should have AO switched off. Also, AO should be off under IDL conditions as IDL generates its own shadowing. Again: AO is for IBL only.
* IDL lighting can be switched off per object, by switching (off) the Light Emitter property of that object. This is worth considering: * for Skin, as Ambient is used sometimes to mimic translucency and subsurface scattering in a fast way, for the older Poser versions. Don’t let your characters be a light source, switch Light Emitter off. * for Hair, as light rays will bounce around forever requiring about infinite render times without adding much to the result. You will then lose the ambient lighting and additional shadowing for that object; it might look a bit flat. Think IBL + AO as an alternative.
* As the environment is supplying a lot of light, either by bouncing direct light around or by adding light from glowing objects (and especially: all-surrounding sky domes), I need far less lights and far lower intensities compared to non-IDL scenes. As a consequence, all the advanced lighting rigs constructed for non-IDL scenes – emulating environmental lighting with lots of direct lights all around, won’t serve very well any more. All I need is a glowing dome for sky, an infinite light for sun, and perhaps one or two spots for support and flash.

Radiosity

Objects which catch light, re-emit light after merging in their own colors, and reducing intensities. This makes a bright red ball warp a reddish light around, noticeable on white floors etcetera. This is the basic principle of Indirect Lighting and served automatically when this lighting mechanism is enabled. The re-emitted light then is used in all further IDL lighting as well, the light rays either die from energy loss or from being captured by the camera (or by being killed by Poser when Bounces is set low).

Light Emitting Objects

In order to use IDL, I need at least one light emitter which sends out rays. This can be a regular direct light, like point, spot or infinite. Such a light is required anyway for creating specular effects and additional shadowing, but for (diffuse) lighting itself I don’t need direct lights at all.

I can make an object glow, by assigning it a high level of Ambient as a material, and it will serve as some lamp immediately. The larger the Ambient_Value, the higher the intensity of the light, the stronger the lamp.

Two balls, one glowing, IDL off, no direct light

Same scene, IDL on

Same scene, IDL, extra direct light on

Same scene, IDL off

Note the strong shadows in the rightmost image, which lack in the third where the white floor is bouncing the direct light, reducing the shadows at the lower back/right side of the white ball. The second image shows shadows from the right ball onto the floor caused by the glow/lighting from the red ball at the left, and also demonstrates the lack of specularity (highlights) in IDL lighting.

Sky Domes

IDL works best when the entire scene is embedded in some kind of enclosure, like a box (walls, floor, ceiling for indoor shots. For outdoor shots, the answer is: a sky dome. I could use a normal (hires, half a) ball at a large scale, but dedicated sky domes have even a larger resolution (more polys to reduce smoothing artifacts) and have their normal pointing inward which generally is not the case with regular objects. And I can apply a texture to the dome to obtain the lighting conditions as were, or could be, present in a real life scene. Large shots from landscape generating software, like Vue, serve pretty well here too.

Note that the Diffuse material channel will “reflect” the regular lighting in the scene, and the Ambient channel will make the dome glow by itself. The latter is the usual response to sunlight, scattering through the atmosphere. The sun itself is best represented by an infinite light, within the dome. Then I raise Ambient_Value to get the proper intensity for this generic atmospheric lighting.

When the sky dome is used for color and intensity of the indirect light, scattering all around, the resolution of its texture map is not an issue. But that leaves the question: is the texture on the sky dome fit for purpose as a background image? Usually, it’s not.

Consider a camera at normal lens settings, that’s 35mm focal length and 40° Field of View (see table below), taking a shot (render) of 2000 pixels wide. The full sky dome, being 360° all around, then would require 360/40 = 9 times my view. And good texturing practices require at least double the resolution of my render. So the sky dome should be assigned a 2x 9x 2000 = 36.000 pixels wide texture, at least. Note that Poser takes 8.192 for max texture size, and you know you’re stuck.

Note that the size of the skydome – or any other 360° environment – does not matter. The Field of View matters, as a shorter focal length (typical for landscapes, say 20mm) increases FoV to 60°, and reduces the required texture to a 2x 360/60 x2000 = 24.000 pixels width.

Focal length (mm)	10	20	30	35	40	60	90	120	180
Field of View (°)	90	60	45	40	30	22,5	16	12	8

So the bets are that you’ll end up with say 8000 pixel wide panoramic image for the skydome, which is too low a resolution for proper background imaging, plus some background image prop holding another 2x 2000 = 4000 pixel wide portion of the high-res version of the panorama just covering the left-to-right edges of the rendered view.

Since this billboard prop might block the skydome lighting considerably (ensure it does not cast shadows, highlights etc) when placed nearby the dome it might need to serve as an active light emitter, the same way the skydome does. When the prop resides at some distance from the dome however this might not be necessary, so you’ll have to test for this a bit.

Next >>

Managing Poser Scenes (15. Atmosphere)

The Poser atmosphere has three aspects: Depth Cue, Volume (both through the Material Room, Atmosphere Node) and Lighting (through the Properties of a Direct Light (Spotlight, eventually Point light). Depth Cue and Volume can be set independently, the Lighting works with the Volume settings. Material Room also has a big button (Wacro): Create Atmosphere. There are various standard options to choose from:

So, let’s take each element apart, and combine them later. Before I dive in: these effects are visible only against objects, reflecting light towards the camera which then is filtered through the atmosphere. Just having a set up a background image and the Poser ground won’t help. You do need a real ground object, and a real backdrop object even when it’s painted black.

Depth Cue

The Atmosphere Node, accessible in the Materials Room, presents for DepthCue: On, Color, StartDist and EndDist.

Depth Cue adjusts the color of objects towards the DepthCue_Color, such that all objects less that StartDist from the camera are not affected, all objects more than EndDist are fully effected and take that specific color only regardless of its materials, and everything in between is effected linearly (so an object at 30% between Start and End gets 30% of the DepthCue Color and 70% of its own.

This reflects the presence of damp or fog, which colors objects slightly towards bluish grey (large outdoor scenes) or to white (real fog, smaller outdoor scenes). It is also a great way to mimic environmental (indirect, IBL) lighting without the rendering costs, for instance sliding colors towards green in a deep forest, and it’s also great for creating under water scenes, coloring towards dark bluish / greenish cyan.

A common trick is the use of “black fog” making objects fade into the dark. Great for evening shots. Or use dark blue, for graveyard and gothic effects. The main thing is: Depth Cue relates to the camera looking into the scene, independent of the lighting.

Thanks to the on/off switch it can be activated independent of other effects to ease setting and evaluating the proper values, and to make atmospheres with volume without depth cue, or the other way around.

Volume

As Depth Cue relates to the camera, so does Volume relate to the lights. Volume effects can be switched on/off themselves too, so they can be set independent of the Depth Cue effects.

The main parameters are Volume Color, and Density. When a direct light illuminates a volume in the scene, that volume acts like a transparent fuzzy object with that specific internal color. The lower the Density the more transparent it seems. On the other hand, each light can have its own Atmospheric Strength parameter:

So some lights can interact more than others. For example:

One infinite white light, Atmospheric Strength as low as 0.000010 plus one white spotlight, angular falloff from 10 to 20, Atmospheric Strengths as high as 0.100. From the different settings of the lights one can discriminate the spotlight from the overall scene lighting. The bluish color is from the Volume settings.

I noted that especially Volume effects take some time to render. A larger stepsize speeds up the calculations at the cost of quality and detail. Increasing the Noise parameter helps to improve on the quality especially at larger stepsizes.

Volume and Depth Cue together

As said: like Depth Cue relates to the camera, so does Volume relate to the lights. But atmospheres of course do both: light rays travel through the atmosphere before they hit an object, and then travel through the atmosphere again to hit the camera. So, let’s add up Depth Cue and Volume:

Which gives me:

The art of making atmospheres now focusses on mixing the proper colors and balancing the other parameters, Volume Density versus Depth Cue Start/End. This happens when I just brighten the Volume Color:

The beam stands out more, but I’ve lost the two balls in the back.

Introducing some structure however (assigning a clouds effects to the Volume):

Given render times, it might be an idea to construct the atmosphere in a simplified version of the scene. Then build the scene with the atmospherics switched off. Ultimately, switch on the atmospherics in the final, detailed scene. From the examples above we learn that we should not spend too much time in tweaking the details of the far away objects.

Standard Atmospheres

The Create Atmosphere Wacro button in Material Room presents four standard settings, as a start for my own:

Fog

Just assigning its own specific Cloud node to an existing Atmosphere node, which does not have any parameters changed.

Smoke

This Wacro changes the atmospheres Volume Color and Desity, and adds a serious set of nodes to both of them. Which does have an interesting effect:

It’s really different, isn’t it? Looks great as a morning fog above the water too, it looks as moving upward.

SmokeyRoom

Does a similar job, except it replaces the Fractal_Sum function by an extended set of nodes, resulting in:

Different structure in the beam of light, this kind of smoke seems to build up, thanks to the ceiling in the room.

Depth Cue

This option leaves all Volume settings as they are, but alters the Depth Cue Color, Start and End parameters. The latter two are determined by the positions of elements in the scene itself. A larger scene gets larger values, quite convenient.

The three Volume choices replace each other when selected, all are independent of Depth Cue. The Depth Cue option adds to either Volume setting.

Next >>

Managing Poser Scenes (16. Background)

Since the Poser virtual world can’t be filled with objects to infinity, I’ve got two ways to define the far away portions

A background shader – this chapter
A defined object with a color or texture (usually a photograph) attached – next chapter

The background shader

As light rays travel from all lights via all objects onto the view plane of the camera, some pixels will hardly, or never, get lit. This is where the “background shader” kicks in, and fills the emptiness. The Poser background shader can be set for the Background notion in the Materials Room. Background is not an object, like the atmospheric Volume is not an object either.

The actual working of the elements is a bit confusing, as you can see there are

The “Current BG Shader” of Background root node (root nodes don’t have an output connector at their upper left)
The BG Color node
The BG Picture and BG Movie node
The Black node

Now I’ve got the Preview and the Render, and the question: which one is showing what?

The Preview is arranged for in the Display menu:

When an image is loaded into the BG Picture node, either by assigning one as the Image_Source parameter or by loading one via the Import \ Background Picture menu option,

the Show Background option in the Display menu becomes available. That is: the BG Picture node should be connected to the Color parameter of the Background node. Then, when the menu option gets checked, the picture is shown in the preview. The image, and hence the content of the Image_Source parameter in the BG Picture node can be deleted by using the Clear background Picture menu option.

A similar scenario holds for displaying a movie in the preview: load one in the Video_Source parameter of the BG Movie node, or import one via the Import \ AVI Footage menu. The Show Background Footage option becomes available and can be checked. Again: the BG Movie node should be connected to the Color parameter of the Background node.

When nothing is checked, or the checked Picture / Movie option is not connected to the Background node, you’ll get the BG Color node contents in the preview, whether it’s connected to the Background node or not.

The Rendering is arranged for in the Render Settings:

The first three options pick the contents of the BG Color, the Black and the BG Picture node, the latter has to be connected to the background node’s Color parameter. The last option: Current BG Shader, picks up whatever is connected to the Color parameter, and multiplies with that color swatch too!

Again:
In Material Room I’ve got the background root node, and four basic nodes: Black, BG Color, BG Picture and BG Movie. I can connect any of these to the Color channel of the Background node.

In the Display menu, I’ve got options like Show Background Picture, Show Background Footage and Use Background Shader Node. Only when the BG Picture node is connected with Background, the Show Background Picture option becomes available to turn showing the background picture in the preview on/off. Only when the BG Movie node is connected with Background, the Show Background Footage option becomes available to turn showing the background movie in the preview on/off. The Use Background Shader Node menu option has not shown any effect on anything up till now. Sorry for that.

From the File menu, I can Import either a background picture or background footage. When importing Background Picture, Poser loads the BG Picture node, connects this node with Background (hence dims the Show Background Footage option) and switches Show Background Picture to ON. When importing Background Footage, Poser loads the BG Movie node, connects this node with Background (hence dims the Show Background Picture option) and switches Show Background Footage to ON.

Do note that I can set the BG Color from the Document panel directly, using the (second) color-swatch option at the bottom-right. So, for handling backgrounds, I don’t have to enter Material Room at all.

In Render Settings, I can select the render background almost independent of my choices for Display, or the node-connections in Material Room. That is: I can render against Black or Color even when the preview is showing Picture or Footage, with Picture / Movie node connected and the Display menu option switched ON. I also can render against Picture or Footage while the preview is not showing it, having the Display menu option switched OFF. But in order to use Picture or Footage in either preview or render or both, the corresponding node must be connected to Background in Material Room. To the Color swatch.

The other way around: how to rotoscope against a movie.

First, I go File > Import > Background Footage. This will load the BG Movie node, connect it to Background en switch ON the Show Background Footage option in the Display menu, so the footage will be visible in preview.

Then, in Render Settings, I have to select Render against: Background Picture (or Current BG Shader), and the footage will appear in the render results as well. That is: provided I use a save / export format without transparency: a series of PNG’s will not show any background anyway!!

Next >>

Managing Poser Scenes (17. Backdrops)

Instead of filling the empty space and non-rendered pixels in the result by a background image, I can put objects in the scene. From simple planar billboards or screens like the backdrops in a real-life photographers studio, walls of a room, to varied setups representing outdoor scenes with more depth. Cycloramas, dioramas, environment balls and more – supported with additional partial billboards and images with alpha channels – all serve the purpose of building a partial environment in the scene.

The one question that comes up every time is: what’s a proper size for images used on those backdrops? Simply stated, the amount of pixels that can be seen on the result should be at least twice (and at most four times) the amount of pixels in the result itself. This has to do with texture sampling and pixel processing statistics, a simple one-to-one ratio might result in loss of quality. As Poser puts a 8192 limit on texture sizes, this implies a 4096 limit on good quality render results – as far as backdrop images are concerned.

And what about full 360° environments like a sky dome?

Consider a camera at normal lens settings, that’s 35mm focal length and 40° Field of View (see table below), taking a shot (render) of say 2000 pixels wide. The full sky dome, 360° all around, then would require 360/40 = 9 times my view. And as good texturing practices require at least double the resolution of my render, the sky dome should be assigned a 2x 9x 2000 = 36.000 pixels wide texture, at least. Note that Poser takes 8.192 for max texture size, and you know you’re stuck. Note that the size of the sky dome – or any other 360° environment – does not matter. The Field of View matters, as a shorter focal length (typical for landscapes, say 20mm) increases FoV to 60°, and reduces the required texture to a 2x 360/60 x2000 = 24.000 pixels width.

Focal length (mm)	10	20	30	35	40	60	90	120	180
Field of View (°)	90	60	45	40	30	22,5	16	12	8

So the bets are that you’ll end up with say 8000 pixel wide panoramic image for the sky dome, which is too low a resolution for proper background imaging, plus some background image prop holding another 2x 2000 = 4000 pixel wide portion of the high-res version of the panorama just covering the left-to-right edges of the rendered view.

Object versus Shader

Using a background object instead of a background shader (picture, footage) does make a difference.

* In order to make proper use of atmospherics, Volume as well as Depth Cue, I do need a background object. Atmospherics don’t show against voids, even not when they are textured using a BG Picture.
* In order to make proper use of Depth of Fields or: focal blur, I do need a background object. The background shader will always be presented sharp, as it replaces empty space. This might give gradually blurring objects against a sharp background, so weird. But of course I can use a blurred background picture for shader, which then remains blurred in renders without Depth of Field set.
* Wherever you turn the camera to, the background shader image will always be the same. Great for stills but not for camera-moving animation.

Not every picture can or should be used for background under all circumstances: it should match the scene, or the other way around. The first issue usually is: brightness, contrast, saturation or: light and color intensities should match. The second issue then is: shadowing. Both issues are best addressed by a complex balance of lighting (position and intensity), materials, sometimes even atmospherics and pre-processing the background image or footage. And please do note that shadows in a background image do suggest the positions of the main lights, so you might have to flip the image to establish a match with the lighting in the scene. And please turn off shadow casting for the background object itself.

Other material aspects “just depend”, usually they are absent. No specular / highlights, no bump let alone displacements, no reflection nor transparency or translucency. But when the background represents a real wall, it just might benefit from specular, highlights and some bump.

Perhaps the backdrop object shouldn’t even respond to Indirect Lighting (switch off its Light Emitter property then), or the other way around: it should emit light from its Ambient channel to compensate for blocking the environmental lighting from a sky dome.

There is no single best way, but perhaps these notes might serve as a checklist. Happy Rendering

Managing Poser Scenes (01. Intro)

Badly lit or rendered images are like vampires: they’d better stay out of the daylight.

Download this tutorial in PDF format (3.5 Mb).

Introduction

Working with Poser is like working in a virtual photographer’s studio. And in order to master the tools of the trade, I enter my empty virtual studio early in the morning, with no models or products to be shot around yet. This leaves me

* the camera, or: the virtual camera body with lenses
* the renderer, or: the virtual camera backplane or film
* the lights
* the atmosphere
* the backgrounds

In this series of tutorials, I’ll discuss them one by one.

Next >>

Creating Hair in Poser (Paula Sanders)

Introduction

A decade ago, Paula Sanders wrote her “Creating Hair in Poser 5” tutorial. She investigated on the styling of medium to long hair, without using the Calculate Dynamics features.

This single webpage tutorial, plus four webpages with hints on creating such hair effectively, can be found here:

http://www.perpetualvisions.com/articles-and-graphics/tutorial-poser5-hair.htm

while also Paula’s other works (www.perpetualvisions.com and http://www.perpetualvisions.com/tocs/toc-articles.htm) are definitely worthwhile looking into. At least they have helped me a lot during my journey through 3D image creation.

As Poser Hair Room is concerned, hair can be styled (turning you and me into a virtual hairdresser), or hair can suffer (or benefit) from dynamics like wind and fierce movements, or both. The latter has its issues, as most women do know from reality: severe dynamics can completely ruin the costly efforts of a talented hairdresser.

Before considering dynamics, it’s good to realize what can be done by styling alone. This is exactly what Paula presents in her perfect tutorial on this, which follows right here. I only can add value in two ways:

Bundle the various webpages into one PDF
Annotate her tutorial and hints, to give some extra depth and background information.

All with Paula’s consent, in case you wonder.

So let’s give the stage to… Paula Sanders. Either check the link above (Paula’s web), or download the annotated PDF (1 Mb).

Case Study: Dressing Up (1 Intro)

Alyson gets dressed, and plays the Marilyn game

Download this tutorial in PDF format (0.5 Mb).

Introduction and Preparation

This Case Study presents some basics for handling the Poser Cloth Room.
I’ll present a Poser girl wearing a simple dress and a scarf, animated to take some poses.

The girl is Alyson, lives in the Poser standard runtime (People \ Alyson section), and comes undressed to this tutorial session. She’ll take the following poses:

T-stand, in frame 1
Reaching01 (People \ Alyson \ Standing), in frame 30
Reaching03M (ditto), in frame 60

While she’s still undressed I step frame by frame through the animation to inspect on clothing issues. Bodyparts should not intersect, and should not tough in places where any cloth has to pass through. Cloth does need some room to maneuver.

The last frame presents some issues with the right arm. The upper arm intersects with a breast and the hand rests upon the knee. Both cases will prevent any future dress to flow gently around the body. On top of that, the shoulder part just looks ugly. From an animation point of view the whole sequence can be improved such that the ball of the right foot stands still, so she’s slightly stepping back in the first part, and bending forward in the second.

After fixing that, I step back to frame 1 and load a dress and a scarf for Alyson (Props > … Clothing > Alyson Clothing > Dynamic). Both are simple props, which fit her neatly. I give the scarf a reddish color to make it stand out.

I save this intermediate file.

Case Study: Dressing Up (2 First Simulation)

Now I enter Cloth Room, click [New Simulation] and enter a name (eg “Alyson”), set the end frame to 90 (1 sec after the last pose), put 15 into Drape frames and click [OK].

In that panel 1 Cloth Simulations I can come back to this settings with the [Simulation Settings…] button.

Next I step into the 2. Cloth Objects panel, click [Clothify…], from the props I select the Dress and confirm with [Clothify].

Then I do the same with the scarf. As a result, when clicking the small triangle in panel 2, both should show up.

In the same panel, I use [Collide Against…], the [Add/Remove] and check Alyson to tell that scarf and dress will collide to her. And to the ground, if I want to. Given the poses that won’t be the case but when she kneels or lies on the ground, it’s required to prevent the cloth from falling through it.

The small triangle should show both, if that’s what I want. Now be sure to check Alyson, and check Ignore head and ignore feet collisions, as the dress will not collide with them anyway.

I’ll skip panel 3 for the moment, I don’t need more details for the clothing, and I also leave all the values in panel 4 as they are. By selecting the scarf and the dress in panel 2, I can see both have slightly different parameter settings in panel 4. These parameters define their physical behavior.

The only thing I’ve got to do now I click [Calculate Simulation]. This shows a progress bar, that first tells me it’s draping (those 15 frames I asked for in the Sim Settings, they take place before frame 1 of the animation), and then simulating (the 60 frames from the animation).

The progress bars tell me that the last finished frame took about 0.6 sec.

Case Study: Dressing Up (3 Analyzing and Improving)

For analyzing the result, I step framewise through the resulting animation.

During frames 6-32 I see the scarf passing through the dress, and around frame 50 the scarf passes through the head, to land on the ground later on.

To repair for that

I open the sim settings in panel 1, and check the Cloth self-collision option (lowest from the three offered). Although dress and scarf are separate objects, they are both clothified in the same sim. That is meant by “self”. Not only dress-to-dress and scarf-to-scarf but dress-to-scarf as well.
Every time there are multiply objects clothified in the same sim, this option should be checked to prevent on passing through another.
I open the collision settings in panel 2, for Alyson. I had checked the Ignore head collisions since I guessed that I could afford that, but I was wrong. So I uncheck this one.
Note that when the scarf lands on the ground someway, I’d better uncheck Ignore feet as well to prevent the scarf from falling though them.

In panel 4, I [Clear Simulation] and [Calculate Simulation] again. In the result I see the dress coming at rest soon after frame 60, the free swaying scarf dangles a bit even at frame 90.

This is my result at frame 75.

Case Study: Dressing Up (4 Girls have more fun)

Now let’s have some more fun. I [Clear Simulation], and back to Pose Room to set the animation cursor in frame 90 and give Alyson an extra pose: Standing02, plus some extra bend through the knees. And I check the added range for unwanted intersections of body parts.

WARNING: in Poser 9 / Pro 2012 the Wind Generator seems to be broken, it won’t affect the cloth sims any more. It’s not you, it’s a software bug.

Back to frame 1, and I choose menu > Object > Create Wind Force. A wind machine drops on the floor, and xRotate=90° makes it point upwards. I move it under her right foot, set the Range to 0.35 (it measures in Poser native Units, 1PNU = 262 cm so 0.35 makes it to her hips) and I set the amplitude to 0. Then I set the animation cursor at frame 75 and click [+] to enter a keyframe, and in frame 90 I set the wind amplitude to 1, while increasing the animation length to 120.

For more precision, I open (menu > Window>) Animation palette, find the Wind Forcefield and its Amplitude and open the Graph.

Back to frame 1 I set this graph to Linear, which guarantees me a wind speed 0 till frame 75, a gradual ramp up till 90 and a constant after that.

Time to save again.

Back to Cloth Room, and I open the Sim Settings to elongate the sim to 120 frames, 30 after the latest pose in the animation. Time to [Calculate Simulation].

Not satisfied? I changed:

The wind amplitude from 1 to 1.5 (more wind)
Selecting the dress in panel 2, the Air Damping from 0.02 to 0.2 (dress responds stronger)
And Fold Resistance from 30 to 3 (dress deforms easier)
The simulation length to 150 frames

and with some altered lighting and a hairdo…