How to Recreate a Dream Graphic with a Character, Photograph, and Caption

To recreate a dream graphic that layers a character, photograph, and caption, record their positions, overlaps, and visible edges first. Keep uncertain details marked before adding illustrative choices.

The task is to make a visual record of those three elements and how they meet. Their presence alone does not establish which sits in front, whether the character belongs inside the photograph, or where the caption appears. A faithful reconstruction preserves those distinctions wherever they are known and leaves unresolved details explicit.

The method below combines a layout sketch with a short record of the relationships between elements. The linked documentation supports the drawing techniques; it does not describe this dream. Any worked drawing choice is labeled as an illustration.

Establish what is visible before choosing a composition

Begin with three separate descriptions: character, photograph, and caption. Describe each only as far as the scene supports. For the character, that might include an outline, pose, orientation, or visible portion. For the photograph, distinguish its outer boundary from the subject pictured within it. For the caption, separate remembered wording from the appearance and position of the text.

Do not assume that the photograph fills the background. It could occupy a bounded area, but even that boundary should be drawn only if it is remembered. Likewise, “caption” identifies a text element without establishing a box, border, font, or location beneath the image.

Use three labels in your working notes: observed, uncertain, and illustrative. An observed detail belongs in the reconstruction. An uncertain detail needs a visible note alongside it. An illustrative choice helps build a drawable version but should remain identifiable as a choice.

For example, choosing a square working canvas is an illustrative setup when the original proportions are unknown. It must not become a claim that the dream graphic was square. The same applies to a substitute character outline or a neutral font used to represent unreadable lettering.

Map the geometry through boundaries and contact points

Draw the outer extent of the composition if you remember it. Then block in the occupied area of each element using simple outlines. At this stage, concentrate on width, height, tilt, spacing, and contact. Detailed rendering can wait until these relationships are settled.

Compare elements directly. Is the character wider than the photograph? Does the caption align with a photographic edge or with the character? Is there a gap between them? These questions produce more useful drawing instructions than a description such as “the character was near the picture.” Record only the relationships you can actually specify.

An optional measuring method is to describe positions as fractions of the working frame. As an illustrative example, a photograph might begin one-quarter of the frame width from the left edge. That fraction makes a drawing reproducible; it is not a measurement of this scene unless the remembered layout supports it. Avoid assigning precise numbers to uncertain distances.

Pay particular attention to the places where boundaries meet. A character outline that continues across the photograph’s edge produces a different arrangement from an outline that stops there. A caption touching the photograph differs from one separated by a visible strip of empty space. Preserve those small junctions before refining color or texture.

If you use a separate visual reference while drawing, keep its proportions under control. Krita’s Reference Images Tool allows references to be moved, resized, and rotated, and its Keep aspect ratio option prevents distortion during resizing. Use such a reference only for the details it genuinely supplies. Krita Manual: Reference Images Tool (https://docs.krita.org/en/reference_manual/tools/reference_images_tool.html)

Record which element covers which

Make a short relationship list for the three pairs: character and photograph, character and caption, photograph and caption. For each pair, note whether they overlap, remain separate, or have an unresolved relationship. Where an overlap exists, identify which visible edge continues and which is interrupted.

This list prevents the names of the elements from silently determining their order. The caption need not be the frontmost element simply because it is text. Nor does a photographic surface automatically sit behind the character. The visible intersections should determine the reconstruction.

In a digital editor, distinguish moving an element across the canvas from changing its position in the layer stack. Adobe documents these as separate operations: the Move tool repositions content, while dragging a layer in the Layers panel changes its stacking order. Correcting one does not necessarily correct the other. Adobe: Reorder and move layers (https://helpx.adobe.com/photoshop/desktop/create-manage-layers/transform-manipulate-layers/reorder-and-move-layers.html)

Keep the three elements separately editable while testing the arrangement. As an illustrative example, if a caption crosses both the character and photograph and stays continuously visible, placing it above both is a useful starting construction. If its letters disappear behind one element, that construction needs revision. Neither arrangement is established by the supplied scene description.

Preserve material relations without adding effects

Separate three questions: where does an element end, what covers it, and can anything be seen through it? They call for different drawing decisions. A missing section of a character might be covered by the photograph; it does not necessarily mean the entire character is translucent.

Opacity changes the transparency of a layer. Adobe explains that lowering overall opacity reveals more of the layers beneath, while blending modes change how pixel values combine. Consequently, reducing opacity across the whole character would introduce a widespread relationship that may not match a single local overlap. Adobe: Set layer opacity and blending modes (https://helpx.adobe.com/photoshop/desktop/create-manage-layers/apply-layer-effects/set-layer-opacity-and-blending-modes.html)

When only a particular area needs to be hidden, a layer mask provides local control. Photoshop’s documented mask operations can reveal a selected area or hide it. This is useful when reconstructing a precise interruption at a photographic boundary without changing the visibility of the entire element. Adobe: Add layer masks (https://helpx.adobe.com/photoshop/desktop/create-masks/layer-masks/add-layer-masks.html)

Treat surface qualities with the same care. A visible white border does not by itself establish thick paper. A sharp character outline does not establish a physical cutout. Add thickness, cast shadows, glossy highlights, softened edges, or transparency only where the scene supports them. Otherwise, keep the drawing simple and note what remains unknown.

Recreate the caption as wording and visual shape

Record any remembered caption wording separately before positioning it in the drawing. Preserve known capitalization, punctuation, and line breaks. If only part is legible, transcribe that portion and describe the unreadable remainder in an accompanying note. Do not finish the sentence merely because a plausible phrase comes to mind.

Then record the caption’s visible footprint: number of lines, approximate width, orientation, alignment, and relationship to nearby boundaries. Even when wording is unavailable, these features can still contribute to a useful visual record.

If the lettering cannot be identified, a neutral font can serve as an explicitly illustrative stand-in. Match only supported properties, such as a narrow line of text or two stacked lines. Keep the stand-in wording out of the reconstruction if no words are known; an annotated outline of the text area may be more faithful.

Check every point where lettering meets another element. Does the caption continue across the photographic edge? Is part of a letter covered by the character? Does the text follow a tilted surface? These are geometric details to retain, independently of what the caption says.

Record changes as separate states when needed

If the graphic changed, preserve each distinct remembered arrangement in sequence. Give each state its own sketch and describe the transition between adjacent states. Record which element moved or changed, what its earlier condition was, and what remained fixed.

Distinguish the sequence of the scene from the order in which you draw it. Building the photograph first and adding a character afterward is a construction workflow. It provides no evidence that the photograph appeared first in the dream.

As an illustrative example, a caption could move while the character and photograph remain fixed. A record of that change would need the caption’s earlier and later positions, plus its overlap relationships in both states. An arrow alone would omit those details. Where the transition itself is unknown, preserve the two known states without inventing an intermediate movement.

If no change is remembered, one composition is sufficient. The presence of layers does not require a transformation sequence.

Finish with a relationship check

Compare the reconstruction with the notes in a fixed order. First check the overall frame and the occupied areas. Next check the three pairwise relationships. Then inspect local boundaries, caption placement, and any recorded changes. Finish by reviewing every illustrative choice.

• Are the character, photograph, and caption represented only to the extent known?

• Do their relative sizes, gaps, alignments, and tilts match the record?

• Is each overlap distinguished from transparency or a local hidden area?

• Are caption wording and line breaks preserved wherever known?

• Does every recorded change retain its earlier and later states?

• Are unresolved details and illustrative substitutions still identifiable?

Keep the working notes with the drawing so that an attractive substitute does not lose its label. The useful result is a reconstruction whose character, photograph, and caption retain their supported relationships, with enough clarity to revise an uncertain edge or position without rebuilding the whole scene.

Sources

• Reorder and move layers — Adobe Photoshop (https://helpx.adobe.com/photoshop/desktop/create-manage-layers/transform-manipulate-layers/reorder-and-move-layers.html)

• Set layer opacity and blending modes — Adobe Photoshop (https://helpx.adobe.com/photoshop/desktop/create-manage-layers/apply-layer-effects/set-layer-opacity-and-blending-modes.html)

• Add layer masks — Adobe Photoshop (https://helpx.adobe.com/photoshop/desktop/create-masks/layer-masks/add-layer-masks.html)

• Reference Images Tool — Krita Manual (https://docs.krita.org/en/reference_manual/tools/reference_images_tool.html)