"Render Selection at render size"

[PhotoComix]

I wonder if would be possible add a option for "Render Selection at render size":
what i have in mind is VERY different from using the selection tool for zooming , it would use the same tool to select but then  will interpret the selection differently

A similar result is possible manually but in a very painful way :
if the the selected "zone of interest "is as example large 1/10 of the image would be needed render everything 10X bigger then desired size, and then crop away the 9/10 of the render

This has obvious disadvantages in rendering time

But i believe a similar process may be MUCH optimized and automatized :

Using the same example of a selected "area of interest" large 1/10 of the image , a virtual image 10X bigger of the final desired size
should be maybe somehow considered BUT only the selected 1/10 has to be rendered,
so maybe is possible avoid to compute the most of rest and surely is possible avoid to render it (with the obvious advantage in rendering time  )

Maybe would be needed some info on the pixels closest to the selected area for some interpolation ?
 
But anyway i suppose it would require much less computation time then rendering a image 10X bigger then needed just to have its 1/10 at desired size and discard the other 9/10

to me happens very often to notice in renders   details fascinating but too tiny and somehow impossible to render big if not as details in really gigantic render

 i really would love a similar option !
(or any other good alternative to render at 16000x12000 to get the desidered part at 1600x1200 )

[MatzeR]

Oh yes. I miss this function, too.
In other render software like Vray or Mentlaray you always have a "render area" option. This is extremely useful when you have a smaller part in your picture that is not satisfying and you want to enhance this part without always rendering the complete image.
I don't know how they do this so that for example the shadow of an object outside this area is correctly seen in the selected area. I think backward tracing - from the pixel to the camera an the light is one point. If this is possible with raymarching I don't know.

[PhotoComix]

I will really love a similar option

[Jesse]

There is a function to 'Recalculate a selection' in the postprocessings, but for zooms you have to navigate towards the object in 3D.
The zoom value in m3d defines the size of the startplane and all values that are related to it, but it is more a 2D relict since you can also do 2d slices in m3d.

[PhotoComix]

There is a function to 'Recalculate a selection' in the postprocessings, but for zooms you have to navigate towards the object in 3D.
The zoom value in m3d defines the size of the startplane and all values that are related to it, but it is more a 2D relict since you can also do 2d slices in m3d.

"recalculate a selection" seems something different what i intended is "recalculate only the selection BUT at render size " that will not use any zoom but create a  gigantic image of which only the selected part has to be (fully) reicalculated

[Pauldelbrot]

So, you don't want the camera moved closer, but instead its field of view narrowed to a specified subset of its previous one, and that field rendered at the full image dimensions?

[PhotoComix]

So, you don't want the camera moved closer,

correct !

but instead its field of view narrowed to a specified subset of its previous one, and that field rendered at the full image dimensions?

Humm ... not exactly , if was that could be already be solved just by changing the FOV value

I try to explain again :
suppose i see a interesting detail large 500 x 500 inside a 3000 x3000 render and i select it
i would like the equivalent of render at 18,000 x18,000 of the whole image so that detail become at 3000x3000 , as the whole original render
 The advantage would be that the part to render will be only the 3000x3000 piece  not the remaing part of this "virtual 18000x18000 image
)

Of course this may take longer then a normal 3000x3000 render , but hopefully would be much quicker then made a 18000x18000 render just to crop from it a 3000x3000 detail :
it would give exactly the same visible result  but will be much quicker

[Pauldelbrot]

That is the same thing, unless you're leaving something important out of your description of what you want to do.

It's just that doing it manually requires turning the camera to center the view on the center of the region you want to draw, and narrowing the FOV setting so that region fills the view, and possibly setting the view aspect ratio different, which isn't nearly as easy, UI-wise, as simply selecting a rectangular area would be.

[PhotoComix]

That is the same thing, unless you're leaving something important out of your description of what you want to do.

It's just that doing it manually requires turning the camera to center the view on the center of the region you want to draw, and narrowing the FOV setting so that region fills the view, and possibly setting the view aspect ratio different, which isn't nearly as easy, UI-wise, as simply selecting a rectangular area would be.

Sorry I can't follow you ...do you think the feature i requested possible and useful , or the opposite , or what ..?

All single periods of your reply are clear ( except the  "setting the aspect ractio different" , what you intend for  "aspect ractio" and how to set it  differently ?) but i don't get the whole

Are you saying that you may explain how to solve manually ?
If i found a example as a m3i + the correspondent  render with boundaries the area i want enlarged  marked  you may show how to solve manually ?

I was talking/requesting a solution for peculiar cases , cases that can't be solved with any of the methods i know

 1) because the object to enlarge instead tend to vanish when zooming or lowering the FOV , and raise the far plane values is not sufficient to prevent the vanishing

AND/OR

2) because if "enlarged" , no matter how "enlarged" if by zooming or changing camera position or changing FOV  , or combinations of the 3 the object changes dramatically shape

AND/OR

3) More rare cases:
i.e because for some reason zooming in or lower the DOF don't make the object closer, neither bigger, till start to fade away




I am really hope for a smarter solutions then enlarge everything , i hate to wait for  long calculation for the pixels that will be crop away  .

So i would really appreciate a new option for :

Run again  last parameters but with image size multiplied as needed to enlarge as required the selected area
Of this new image compute ONLY what needed to render the selected area

[Jesse]

Just wanted to let you know that you can spend your effort for other things unless you are very rich and pay me for such features that could be done more or less already by the included functions 

You want to point the viewing direction in the navi towards your point of interest and lower the FOV until you get what you want...
or just step towards your POI like the usual way in 3d.

 

[taurus]

Just wanted to let you know that you can spend your effort for other things unless you are very rich and pay me for such features that could be done more or less already by the included functions