I find the fact that Softimage often makes me doubt my own sanity by saying that the scene I'm about to close has never been saved, even though I saved it just two clicks earlier.
Also having just launched Softimage, and gone to the file menu to open a scene, I find it very annoying to have to confirm that I don't want to save the completely empty scene the I'm presented with at start up. This is especially annoying when you are trying to open a scene file that you know will take a good few minutes to open. So you head off to make a nice cuppa only to return and find that the software hasn't even started opening the scene file yet as you still haven't confirmed that you don't want to save an empty scene. Grrrr!
One more thing,
the inconsistent file browsers. The windows 7 file browser often seems to be standard browser for functions inside softimage. ( browsing to the out put path of a cache for example.) It would be better if all the file browsers were the standard xsi browser.
+1 to that one !
Here’s my short list, some dating back to v1.x:
1) Inspecting a mixer source or mixer clip via the RMB context menu displays a collapsed PPG that must be manually resized via dragging the mouse to see it’s contents.
2) Transfer / Merge Attributes editor found in the Poly.Mesh merge tool, GATOR, and a host of other tools only shows top two rows in the upper pane of the dialog and is NOT resizeable.
3) Transfer / Merge Attributes editor defaults to pairing up cluster properties (UVs, vertex color, user normals) by chronological order of creation instead of by name.
4) Updating a mixer source does not update mixer clips pointing to it from the mixer timeline.
5) No easily accessible UI for looking at image sources/clips on a mixer timeline. Must navigate to the image clip folder in the scene explorer and know to select the image clip(s) to see anything at all. Images should be an available option from the mixer.
6) No SDK access to Spreadsheet view.
7) No option for scaling objects on axes other than local (such as global, object, or COG)
8) Cannot adjust explorer view filter to only show custom properties or other property type data without showing the rest of the object’s structure.
9) Deleting selected object from the scene when in an explorer set to “selection” scope causes scene explorer to go blank even when it’s locked (auto refresh deactivated). Really? Can’t it just maintain the current view?
10) Texture Editor relax tool doesn’t relax.
11) No place to hide custom userdata that isn’t in plain view of the user. Would be very useful for custom tools.
12) No paint UI or ability in SDK to create a useful 3D paint system.
13) Performance very bad with many objects in scene regardless of object complexity.
14) No ability to auto update self installing custom properties in a scene after code changes.
15) No falloff icon/manipulator for point lights
16) No ability to send events/callbacks to netview for custom tools to respond (example: on selection change, would like a netview based tool to respond)
17) No ability to preserve work on NURBS Surface when converting to polygon mesh….other than trivial geometry.
18) Realtime shader system overhauls that shows vertical improvement instead of lateral.
19) Lack of UI customization throughout. Makes custom tool development more difficult than it should be.
20) Ability to substitute operators/operator inputs in the construction history outside of text and curves.
21) Ability to read/write the construction history from the SDK other than a trivial XML dump. What’s needed is low level access to operator inputs/outputs so the operators can be recreated exactly using custom importers/exporters. Example: the ‘movecomponent’ operator does not expose any parameters. Therefore a custom tool which needs to build an object from a file description cannot recreate the movecomponent operator on the generated object’s stack.
22) ASCII file format for all presets, scene files, models, and other externally referenced data specific to the application.
23) A finished NURBS modeling toolset.
24) Ability to read / write NURBS data in parameterizations other than Non-Uniform.
25) Ability to get access to a reference model’s Delta via the scripting object model.
26) Ability to write to a PointCloud from scripting. Example: apply color values to the points in the point cloud.
27) Ability to duplicate an object contained in a referenced model. That is, the resulting duplicate should be local to the scene and have no affiliation with the referenced model from which it was derived.
28) Ability to define presets / default values for commonly used tools.
29) An Image clip viewer available directly from the viewport menus which allows panning, zooming, resizing, channel component viewing, etc… that was written in the last 15 years.
30) Tools for selecting subcomponents from samples (eg: uv --> polygons). Needed outside the texture editor.
31) Customizable context menus for tools allowing removal and rearranging of tools, and allows arbitrary ordering/placement of custom tools.
32) Tear away controls from any toolbar that doesn’t require an entire package of buttons/controls to be torn away with it. Example: I just want to tear the snapshot button off the FCurve editor toolbar. I don’t want the swap buffer, HLE, and other tools to come with it.
33) A toolbar/layout/shelf editor that was built in the last 15 years and works on drag n’ drop and tear away editing principles. Heck, even the old SI3D dialog editor was more intuitive than the XSI version.
34) Ability to put expressions, constraints, and other data types into mixer sources via the SDK, AND the ability to do so without requiring the parameters contained in a mixer source to exist in the current scene.
35) Why do primitive NURBS circles and squares default to having so many control points?
36) Using any tool which creates a sample clusterproperty (UV, vertex color, user normals, …) results in the newly created property being inserted into the most recently created sample cluster. If you work in a system which requires clusters to be named (oh, I don’t know, like a games development workflow) you cannot control how data is generated resulting in a lot of misplaced data breaking stuff.
37) Value locks that actually work would be appreciated when used in referenced models.
38) When transforming subcomponents individually in local mode, there should be a ‘center’ icon drawn for each subcomponent so the user can see where each subcomponent will move.
39) subcomponents should transform relative to the selection’s center when working in symmetry mode. Example: a character in a T-pose. If a ring of polygons around the wrist are selected, rotating the polygons when COG and Local are active should result in the polygons rotating around the wrist, not the global origin.
40) Shadow icon colors should show up in the schematic, scene explorer and other views.
41) There should be an easy way to view the R, G, B, and A channels in the viewports. Important when working with vertex colors and other metadata which isn’t necessarily used in the traditional context of beauty pass color. Example: normal maps where the unused blue channel is used for another purpose. Sometimes we need to see that blue channel on it’s own.
42) Ability to work on multiple texture projections simultaneously in the Texture editor without having to lock any.
43) No tools available to convert instances to objects and vice-versa. Same for clones. No tools exist for finding/selecting the masters of an instance or clone. I’ve written my own, but it’s annoying I have to do so.
44) ICE Modeling is restricted to the modeling marker within the construction history.
45) XRay mode is per object. Doesn’t take occluded objects into account for doing useful things like making translucent image planes for roughing out the modeling or texturing of new assets.
46) sliders in the KP / L are lmitied to on/off states and cannot set to other values. Example: a parameter defined as siFloat with range [0…1]. In a PPG this shows up as a slider which can be set to any value in the range, but when this same parameter is edited in the KP / L, the user is restricted to 0 or 1 as accepted values when dragging the slider. Must manually type values to get anything inbetween.
47) Cannot view an animation layer without the base layer. There are times users want to only see the upper layers in isolation so they can see only what has been changed beyond the base layer.
48) Snap to texel is missing from 3D snap toolset.
49) Proportional modeling settings apply globally, but how the values are interpreted by tools are different depending on the view which the feature is used. Example: in the 3d viewports the proportional modeling settings are defined in SI units, but in the texture editor they’re defined as UV values. The problem is falloff as used in the 3D viewports is usually on a fairly large scale (several units) as objects are usually modeled fairly large. But UV editing is usually in range [0….1], so any proportional modeling adjustment defined for the 3D viewports will have no effect in the texture editor because the falloff radius will be larger than the entire UV space which texture editing takes place. Reducing the falloff to satisfy the texture editor results in the falloff being too small for use in the 3D viewports. ARGH!
50) Anytime those nested layers want to work is fine by us. Especially with referenced models.
Is the new dev team making a list out of this? This list is gold. I'd absolutely love to see this kind of stuff addressed in 2014.
On Aug 18, 2012, at 12:03 AM, Eugen Sares <soft...@keyvis.at> wrote:As always, when people have different needs -> checkbox in the RT preferences.
Am 18.08.2012 08:17, schrieb Ciaran Moloney:
I'd rather it didn't do that to be honest.
On Fri, Aug 17, 2012 at 2:56 PM, Eugen Sares <soft...@keyvis.at> wrote:
Am 17.08.2012 15:42, schrieb Jeff McFall:+1
If the render tree window is open and one assigns a new material to the selected object the window does not automatically refresh to show the new material
Not sure if this is intent or not but I often end up changing the wrong material because I expected my new material to be there
Even asked VRay support two days ago if this is related to the connection plugin, but no, it's a SI bug.
I find that when there is a point cloud with a heavy instance shape node attached, viewport interactivity takes a major hit even with the point cloud and shapes hidden.
It is so bad I can't interactively create curves, however (again the point clouds and shapes models are hidden) if i detach the instance nodes everything is fine.
Some time ago (around XSI 7.5 or so) I logged all the settings I did and saved them out as a script.
Now when I have a fresh install, I just run that script.
(the script obviously got some lines added over time and Soft versions…)
- - -
Juan Brockhaus | Head of CG Studio
Prime Focus London Commercials
T: +44 (0)20 7565 1000 | M: +44
(0)7944 720 557
A: 37 Dean Street, London, W1D 4PT, UK
I just export my
preferences and re-import for a new installation or user profile...right click
on the preferences and export, I know it doesnt really export all preferences
like e.g. transform preferences but its something