I have a scene with a moving object and I’m using mblur.
But what I really want to do is freeze time and have the camera scan around the object. But I want it to keep the motionblur of the moving object and I want the particle systems to remain as they are as well… Also the camera is moving along rather quick in relation to the background. which is a motion I want to keep in there as well. And really I can’t go and render like 200 sepparate images cause a single one will probably already take a few hours.
you want the motion blur of as if the object was moving
and you want the motion blur of the background
yeah, it’s possible
you would set the speed (iirc) ipo for the object(s) you want to freeze to move along that one frame repeatedly, and then just animate the rest as you like
the problem would be getting that one object to have that in their speed ipo, while having a good transition into and out of it. it is just about trivial if you could make this it’s own ipo and simply render it out seperately.
If you want to keep the blur consistant and not look like everything stopped, that’s probably what you’ll have to do.
1- Make a path where you want the camera to go (during your stop motion part)
2- Parent the camera to the path
3- Render a single frame and save
4- Offset the camera 1 frame forward (with TimeOffset or the Offset parameter of the constraint if you are using FollowPath).
5- Repeat 3 and 4 until the camera is at the end of the path.
it seems to work when I animate everything… then use the time IPO to freeze the object’s motion at a certain frame and do the round scanning camera after that.
Somthing odd though. It looks as if the mblur uses the frames in backwards order… so if I render frame 5 it ues frames 5,6,7,8,9, and so worth… whilst I think it should be 1,2,3,4,5 if you get what I mean?
This is something I worked out a long time ago for another thread (Aug 2002) Don’t know if it’s the best solution and if it’s still up to date.
Could be a bit of a problem with the particle system.
Hi, here is my improved solution for the matrix effect (stop motion ?), that should work quite well.
-Set up your scene with animation
-Make in total three copies
-Render the complete length of the first copy with motion blur but without the moving objects
(Background Clip)
-Render the complete lenght of the second copy but without motion blur and without any background or
decoration. (BlackWhite clip)
Background color has to be set here to black, the color of the moving objects has to be set to bright white
The result should be a clip in black and white without any gray or color
(This clip is needed as a mask in the sequence editor)
-Render the complete length of the third copy without motion blur and without any background (black)
or decoration, but this time with normal color and texturing of the moving objects (Object clip).
This three clips go to the sequence editor (Shift+F8)
-From Add sequence menu (Shift+A) select Movie and place the Background Clip in chanel 1.
-Add BlackWhite Clip to chanel 2, right above the Background Clip.
-First select Background Clip, hold Shift and select BlackWhite Clip as second.
-From Add sequence menu select Sub (SUB 1-2).
If you now render the sequence (press Do Sequence in Display button and then Anim)
you should see your scene with black holes were the moving objects should be.
If this is the case, merge the two clips and the SUB sequence by selecting all three and typing M (Make Meta).
(merged sequence should go to chanel 1)
-Now, add Object Clip to chanel 2 right above the merged sequence.
-Select both sequences and from the Add sequence menu choose Add.
Rendering of this sequences should give a nice matrix like effect.
only took a couple hours
(not bad considering I am no modeler or animator, lighiting turned out surpizingly good too, and this computer is one of those with the intel cards that crash when you select things sometimes…)
Yes, that would work, but you won’t have the blur like you would have had in motion.
Somthing odd though. It looks as if the mblur uses the frames in backwards order… so if I render frame 5 it ues frames 5,6,7,8,9, and so worth… whilst I think it should be 1,2,3,4,5 if you get what I mean?
AFAIK, it’s always been like that. It’s actually more logical and practical that way (you’d have to add keyframe in negative frame numbers if you’d want correct blur in frame 1).
This is the result of a simple test… there’s a stationary floor with a line of cubes.
Another Cube moves along this line. It’s animated but frozen mid anim using a time IPO. An empty is child of this cube, and the camera is the empty’s child. The empty rotates.
See how it still looks like the cube and camera are moving forwards along that line… and this whilst actually they’re not moving but frozen in time.
This is a lot simpler than anything else I’ve read here so far and works fine.
And theeth… I’m not talking about frame 1… that’s just a forinstance… it just looks like the mblur looks forward in stead of backward.
So I am rendering frame 50 and the image looks like it’s built up out images 50 50.1 50.2 50.3 and so forth… and in stead I’d think it should be 50 49.9 49.8 49.7 and so forth.
If you use the frames before instead of the frames after, that would means you’d need the frames numbered 0.8, 0.6, 0.4 and so forth to render frame 1.
Technically, both solutions are possible and for both solution, you can adapt your work to make it act like the other solution. Somewhere along the line, someone took the decisions to move forward when rendering motion blur. The difference between the two choices is merely technical.
Yeah well… I understand… thanks for the explanation, but I still think the way it works now is utterly wrong. (personal view). It just doesn’t make any sense anim/image wise.
So now I’ll have to animate backwards… ouch. can that even be done with a particle system? (time IPO prolly yes)
I thought about it another second and really honestly… of course for image 1 it should look at 0.9 and so on… and of course some people are going to be confused by that when their image 1 isn’t blurred like they’d like but it makes simple and pure sense.
I just can’t get my head round the decision of making an image blur with future instances in stead of past…
Whomever made that decision must have had an incredibly dodgy moment at the time. (sorry if I’m sounding a bit blunt but really I just don’t get it… it’s such a simply wrong thing to decide).
Maybe it’s just me, but I really don’t understand how either of the solutions can be wrong. The motion blur technique Blender uses work by rendering samples over a period of time. The choice of rendering the periods of time starting at the current frame make sense if you look at it like a photographic camera. The shutter opens on the whole values (frame 1, 2, 3, …) and closes shortely after that. The Bf can directly lead to the shutter speed with some simple calculations.
Well it may just be me… When I think of mblur I think of an effect almost like the mousepointer trail you can switch on in windowz but captured in an image.
I rather believe that the mblur function in blender hasn’t been worked out too well anyway… Honestly having it linked to the OSA is a bit wonky and it would be nice to have settings like a: a setting for how many images to capture for mblur and b: over how long a time these should be stretched (which can be a positive or negative nr) and maybe c: a gradient chooser… whether things run exponential or not. then d: the intensity thing… and last e: an on/off switch.
I think without those things it’ll never really be fully functional
Just did a simple test with a cube moving 25 frames. I want to render the cube as it appears at frame 15. Do it first with no motion blur and note where the cube is. Switch on motion blur and the blur extends forward from where the cube physically is at frame 15.
That’s impossible. I should not be able to see the future in motion blur, only the past.
Now if we are simulating shutter speed and film speed and all, then technically, frame 15 should be the moment the shutter slammed shut. ie, the last image captured on the film.
It’s like Blender is saying that with Motion Blur on, you can’t pinpoint the object at the frame you are rendering, you must instead see several frames into the future. I think it would be more logical for Blender to show you the frame you want to render as the most recent, plus a few frames leading up to that point for the blur.
That would be a problem rendering around frame one as has been stated.
Why should it “technically” be the frame when the shutter is shut? It’s as logical that the current frame is the frame when the shutter is open.
Stupid exemple to illustrate my point:
A canon is shooting red balls one after the other. You decide to take a picture of a ball as it exits the canon (frame number X). What will you see in the picture?
The ball is inside the canon and you can only see it blurred a bit throug the opening.
The ball is outside the canon and you can see it blurred as it gets away from the canon.
Which of the can you describe as “Taking a picture as the ball exits the canon (frame number X).”?
I see your point but I’m going to labour mine anyway
The frame number represents an exact moment in time. I’m a PAL kind of a guy so frame 26 represents a point in time exactly one second after the animation started. Let us say I fire a cannon at frame 1 and the cannon propells the cannon ball with such a velocity that the ball exits the cannon one second after firing.
If I look at the wireframe of frame 25 I see the cannon ball still in the mouth of the cannon, hidden from my view. At frame 25 my cannon ball has not yet left the mouth of the cannon.
HOWEVER, if I render frame 25 with motion blur, what do I see? I see the cannon ball already out of the cannon’s mouth and hurtling toward it’s target. This is a physical impossibility and Blender is not rendering the moment of time represented by frame 25 correctly. Blender has crystal balled frame 25 into the future. We are actually seeing (in the render) reality as it is (in the model) at frame 26 or 27 (or more).
Lets say our cannon is made of crystal and we can see through it. What I want to see when I render frame 25 is my cannon ball, just inside the mouth of the cannon as it is in the modeler, but with a blur extending back down the cannon’s throat.
ie I want to be able to rely on the position of the cannon ball being the same in both model and render at frame 25. That way I can plan animation based on the where things are in the model and not where they appear to be in the render.