Using Background Pics for reference

The other day i was modelling and was meeting a friend so i saved and quit and when i got back, the background picture wasn’t there when i loaded it up again. I have tried reloading the image and i have also tried different images. I noticed that when you use a background and you rotate around the 3d window (using the middle mouse or ALT-right mouse), it disappears but i also tried all the views like side top front etc… Also the picture i am trying to use is scaled to size of the camera lens and i can see it viewing through the camera but it would be a pain in the ass to scale the camera to the right size and position it well to use as a reference, so if anybody knows where my precious background pic has gone, please help me out! I’m stuck half way through a model with no reference!

Also I don’t have a Z plane this could be part of the problem. ALso i don’t know if I got all the terms correct being new so if you don’t understand, just ask.

Ooops :expressionless: It was in perspective mode…
Ok problem fixed now…
wow i feel like a right tool
could a mod delete this or something?


Im using a blueprints for references. So my question is, when I have two pic:
624340 (front side), and 340624(top view) When I place as background there are not the same size. The button ‘size’ I have set up to 10.0 for the two case.

So my question is how can I make the same resolution the two backround pic?(ex: 72 dpi)

For a workaround I have created a 640*640 white pic in Gimp, and I have placed at. So its the same, but Its really hard to center the images.

Another problem with the background pics:
My favorite windows setting are: a big window(left) and three small views at the right(camera(the smallest), front, side). When I setting at all the correspondant background image(top, front, side) it looks good. In the tiny windows are positioned too the background pics.

Now when I change the view in my primary window(the biggest) also the background not switch the corresponding. Not a problem! I change the pic(I use the menu for that), but it must repositioning the image!
And thats is really slow. I must repositioning the images all my view changes …

So how can I parenting a pic for a view point?

I hope you understand my little problem(if not, I can post some screenshot)

Best regards,

Add a cube and scale it big so your camera, lights and objects fit inside it. UV map your blueprints to the inside walls.


Flight %, That´s some ingenous, smart and good idea! I always stumbled on the before (BG not being view dependent). I used to have mapped planes as references in other apps (Maya), because of speed advantage over BG image, but never used that in Blender…

Khiraly, you´d better do the alignment in Gimp. And you should aldo use 16 power, square sized images, as they perform faster in Open GL (eg. 64x64, 128x128, 512x512, 1024x1024)

Alexandre Rangel

Thx for the response!

I will try the cube-things.:wink:
In the tutorial origine use a cube too.

But If I think good, is somewhere a bug in blender …
Here is a pic, as you can see the two background image has the same size(710492). For the background image(at right) I have cutted out and rotated in gimp, and place at an 710492 white image.

But the two background size is not the same …
So its a shift(dislocation) for the two view(top and side).
Here is the pic:

Here is the two background pics:

I have made 4 plans(top, front, back, side), I have UV textured. Its look good.

I have made first the front plans, assigned an UV map. After that I have copied(SHIFT-D) and move it for the back plans. I have tried to assign an other UC map, but it change the front plan!
Every time when I try to change the UV map it change another plan too!

How can I make two different UV map? I have tried the Ooops schematic window, but I never know how it works …:frowning:

Thx in advance for any help!


When you make a duplicate [even SHIFT-D], stuff like material, texture or ipo are not duplicated, they are linked! Thus your two planes all link to the same material and texture, if you change one they all change.

You can see that a data block such as material, texture, etc are linked to several elements thanks to the number at the right of the name of your data block. For eg make a plane, then assign a new material to it. Now duplicate your plane, and have a look at the name of your material, there is a “2” showing up at the right of this name. If you click on that “2”, there’s the option “single user”, this means that your currently selected plane will have it’s own copy of this material and no more share it with the other.

Ok same thing, but with the texture:

  • make a plane
  • assign a material to it
  • duplicate the plane
    => see the material has 2 users
  • add a texture to your material
    => the texture has only one user
  • select one plane
  • make the material of this plane single user
    => the texture has now 2 users!

Understand how it works? :slight_smile:

So you have to make sure that the material, and the texture are single users…


Just to avoid confusion:
Plan = tech drawing viewed from above
PlanE = Blender primitive

Also, when you use a new plane and go to the image (UV) window, hit A to deselect all selected faces in face mode before you reassign new faces for the projection, but I’m pretty sure Diz has the right answer to your problem.


Thank for the quick help! Now its more clear.
I have made the 4 plan for the car, its look good.

I have some problem with this method:
The reference image its appear only when Im in mode TEXTURED, but when Im in mode textured the car(what Im building) its not transparent! So I cannot look through of my car. Its really confusing.
Is it possible to have ONE specific objet to be in mode textured and the others in mode wireframe?
Or how you use this technic?

Other problem:
When I look at close on my objet the reference pic is blurry … I like more when its pixeled(like in the gimp) its more easy to work with a pic pixeled as a pic grimy…

Thx in advance,

Yes, for each object, you can choose the “maximal” quality in which it will be displayed. This is done in the Object buttons [F7] => Draw pannel => Drawtype => Wire, to have your object displayed in wires. So do that for the objects you are modelling on, not for the planes you use as reference. :slight_smile:

I really don’t know if that’s possible… I think it can be done for rendering, because in the texture buttons there are options such as InterPol, MipMap, and so on. But these options have no effet on the in-blender display. Sorry, can’t help for that.


Well I searched at all possible places where such an option coud be, but I didn’t find anything, sorry! This is one advantage of the traditional background picture method, that doesn’t interpolate textures.

That could be added to the feature requests if there is really no such option…


You can make the planes with the reference pics on them more transparent by making the ‘Alpha’ value lower.

For the blurry effect of the zoomed background pic: This is because the pic you used is (probably a .jpg) and has a low DPI (dots per inch) value. You can change this in a paint program (save it to 600 dpi instead of 90). The only problem is that the pixels of the original will still be squares when converted. A workaround is to repaint the pic on another layer using the clone tool and save only the new layer at the higher DPI. It’s just a partial workaround as it still captures the pixels.


Just changing the DPI value is does not work to me.
Other size if I do a pic that the size for ex. 4* biger as the original, so it slowly my blender no?

Thx for the reponses. Maybe its really would be good to make it a feature request. But is not a ,feature’’ of openGL? (all the openGLed games, when you look to close at an objet in the game, its blurry too)


If you oversample your picture in gimp, but don’t want to have an interpolated image [blurry as you say] you have to do this:

  • Gimp 2: In the image resize dialog, at the bottom choose interpolation type to none instead of cubic or linear.
  • Gimp 1.2: You have to go to the file menu => preferences => Environment => interpolation type => nearest neighbor.

This will not make the effect dissapear, but it should be reduced. This will also make blender a little slower, but depending of the size of your image and the complexity of your model, the speed may still be fast enough to comfortably work.

To enhance the speed in openGL with textures, always take pictures that are powers of 2 [2^x], such as 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, etc. This really makes a difference. In fact I suspect blender of always taking the superiour power of 2 to display a picture that has any size. I mean that if you give him an image of 800x400 I suspect him of creating a 1024x512 big image, and leave unused space invisible.

Sure, that’s a feature of openGL, but there is probably a way to disactivate it… And if there is, this option should be added to blender, since this would be really useful.

How can feature requests be done? Do we really need to subscribe to the functionality mailing list, or is there a more “light” way to do suggestions?


In the Forum->INterface&tools can you make your suggestion.
Here is the link:

Thx for the gimp instruction!

If you write your specification I post willingly at blenderForum.

I dont like to post , because Im poor in english, and (as you can read) I cant me good express, so the first 4-5 post is nothing else just more explication from me …

I have a big idea that would be good when it exist. Its just a dream, so I dont want to open a new topic(and as I said before its a ,dream’’ (or idea))

So my idea:

The people at this forum have really big problem with the time of the rendering …
My idea, it would be good to make a network-rendering over the internet. Some people have at least a ADSL connection, so I suppose it would be enough.
Many people execute at your home computer [email protected] program, its just would be made similaire.
No to have days of render!:wink: Many people would be capable to create a short film …

To not flooding the network it must be made a credit system. So if my computer have rendered for the network for ex. 1 hours, so it could be give a job(for the network) 10 hours rendered time. (reference par ex. a 1Ghz athlon.)

What you think about?
Is it a good idea, or it must be rest a dream?:wink:

Wow… this would be amazing!! Yes, this is worth trying to launch some synergies, a nice little spec that precises things and that shows the potential possibilites, thus giving to the others the lust of pushing it on. I don’t know if anybody knows something abut verse, but perhaps this technology could allow do to something like that? I’ll try to read the verse page again and that time try to understand at least something, lol!

Yes, I definitively have to create a small wish list about little things that were discussed lately in this forum, and post all that on, and on the feature request tracker that is activated again. But this little list should be separated from the network rendering idea, otherwise things will get too confuse.

I don’t have much time today… perhaps this evening [europe]. But my english is very poor too, french is my mother tongue. Don’t be ashamed of your english, we are many english newbies on these forums. :slight_smile:


If i know good , the dr queue tool is it an already network rendering tool. It must made just for the web. Also made a page around(where the people can register) and put a system acceptable for everyone how will be contribute. AT my home my computer run everytime. And I have at the night many CPU-time unusable …

I don’t have much time today… perhaps this evening [europe]. But my english is very poor too, french is my mother tongue.

I never studied my english:-( The english what I have is it from many linux-manual(rtfm), and tutorials. So my grammaire is really poor. Im not clear with some base things(conditional and so on)

Just en parentheses, moi aussi je parle mieux en francais, meme si c’est pas ma langue maternelle. Alors si cest plus clair on peut discuter en francais.
J’ai pleine d’idees mais je ne me peux pas bien exprimer en anglais. Pour ce reseaux-render aussi je pleine ,specifications’’ alors si tu t’interesses bien alors on le peut discuter et apres on peut ecrire a la forum de blender.

En francias j’ecris l’idee ce que il faut bien tenir en compte.

Il y a deux choses a bien comprendre. Une fois il est interdit a permettre que quelqu’un exploite cette system. A quoi je pense? Premierement a des entreprises, eux il faut acheter des render-farms et peut-ils bien tomber dans le faut, que cette un systeme beaucoup moins chere et comme-ca ils achetent pas leurs render-farm, alors il donne plutot des boulots a notre reseau. Cest rien d’autre qu’exploiter notre network. Nous voulons pas a travailler pour faire des profits des entreprises inconnues.
Ce pour cela il faut introduire une systeme de credite, ou on decide que si quelqu’un investe des CPU-time a notre network alors combien CPU-time peut-il recevoir. A cette point la il faut bien trouver l’equilibre. Cest vraiment importent, on peut dire que cest la cle. Si on rate a cette point-la, les gens en perdent l’envie a l’utiliser.

La deuxieme chose est que moi aussi je veux pas avoir un render-farm. Alors je ne veux pas renderer un film entier. Alors je veux simplement reordonner l’utilisation (maximale) de mon ordi. Alors quand je ne fais pas de render pendant deux semaines est apres je ne me veux pas embeter d’ attendre des jours pour renderer une petite animation.
Pour mieux organiser ca, pendant des nuits quand mon ordi ne fait pas de render alors je voudrais offrir mes CPU-time mais en echange de cette cadaux quand moi j’ai besoin vite des CPU-time alors le resaux doit me donner de CPU-time.

J’imagine comme-ca mon idee, cest mon concept.

Pour la technique. J’ai aussi reflechir a cette chose-la. Comme maintenant il y a plusieurs possibilitees de renderer en reseaux, alors arriere plan de la technique est donnee. Il faut simplement faire capable pour la grande reseaux(internet).

Dr.queue si je sais bien est un render opensource pour blender. Alors la plus simple methode de developper cette logiciel a deja fait.

Quelle parametre de technique devrait-il developper? ALors la premier chose de faire capable de communiquer a l’internet. Deuxieme chose il faut reduire pour le minimum d’echange de donnes. Alors il faut faire minimal de bandwith obligatoire. (il envoi des package a calculer.)
Alors il faut faire que la minimum bandwith soit pas plus grand que la bandwith d’un ADSL.

Deuxieme chose il faut faire capable d’indentifier lui-meme avec une serveur centrale, qui a la systeme credits. (pour ca il faut faire une moteur de portaille(php/perl))
Une fois il a envoye leur identification au serveur donne a le client que combien credit il en a. Cette point-la est aussi importent. Il doit enregistrer les points de credits en serveur est pas chez les clients. Comme-ca on peut echapper des floodes de la serveur est les tromperies.

Pour le debut cest une petite reseau. Alors il faut pas penser a des grandes choses et realisations comme une network d’edonkey ou kazaa.

Alors qu’est-ce que doit-il encore enregistrer au serveur?
Les IPs des client qui ont des boulot pour la reseau, est la quantite aussi(par ex. 10 hours CPU-time) Alors dans le serveur on a des informations des clients des suivantes choses:

  • son nombre de credit(l’argent si tu comprends mieux comme-ca. Alors ca dinimue si on donne des boulots a la reseau et ca augmente si on fait des boulot pour la reseau)

combien heures de travaille a pour la reseaux(le mimimum est 0, est ca veut dire que il peut accepter des boulots)

En gros ces mes idees. C’etait beaucoup plus simple a t’expliquer en francais a la place de l’anglais.

Alors si tu trouves bien cette idee-la on pourrait reflechir ou il faudrait continuer l’echanger des idees. (a ou ici a l’elysiun)

Et je te voudrais aussi demander que si tu trouves bien ces idees alors sil te plait traduit en anglais.

@+ Khiraly