Somewhere between a viral tweet and a YouTube thumbnail, someone invented the phrase “Midjourney batch consistency mode” — and now it’s all over tutorials promising you can auto-generate 30-slide carousels with a single click. Spoiler: that feature does not exist. Midjourney V7 is genuinely powerful, the consistency tools are real, and yes, you can build stunning carousel sequences — but the workflow looks nothing like those tutorials describe.
This guide covers what actually works: using --cref, style references, and disciplined prompt architecture to generate 10–30 visually locked images, then moving them into Later or Buffer for scheduling. No fantasy APIs, no made-up features. Just the real method, which — once you know it — is faster than you’d expect.
By the end of this tutorial you’ll have a repeatable workflow for generating Instagram carousel sequences with consistent visual style and character across slides, organized and ready to upload to a scheduling platform. The consistency won’t be pixel-perfect magic — it’ll be disciplined, structured, and good enough that followers assume you hired a design team.
You’ll need a Midjourney subscription — the Standard plan at $30/month is the minimum worth considering here, since generating 10–30 images per carousel will eat through Fast GPU hours quickly on the Basic tier ($10/month). If you’re doing this regularly, the Pro plan at $120/month gives you more headroom. For scheduling, Later’s free plan works for light use, but paid tiers starting at $15/month add bulk upload and link-in-bio features worth having. Buffer’s free plan is also functional; paid tiers start at $5/month. You’ll also want a folder system — locally or in Google Drive — because Midjourney has no native export pipeline to either platform. Everything goes through manual download or, for the technically inclined, the Midjourney API with custom middleware.
Note 💡
Midjourney’s API exists and supports programmatic image generation, but it has rate limits tied to your subscription tier and no documented endpoint for carousel sequencing. If you want full automation, you’re looking at custom scripts connecting the Midjourney API to Later or Buffer’s APIs — real engineering work, not a settings toggle.
Every carousel sequence needs a visual anchor — the first image you generate that defines the style, color palette, and mood for everything that follows. This is not a throwaway step. Spend time here because every subsequent image will reference this one.
Start with a highly detailed prompt that bakes in your visual rules explicitly. Don’t rely on Midjourney to infer what “cohesive” means — tell it exactly what you want.
flat design infographic illustration, woman in coral blazer explaining financial concepts, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, 9:16 aspect ratio --ar 9:16 --v 7 --style rawRun this prompt and upscale the result you like most. Save that image — it becomes your style reference for every other slide. The more specific your prompt, the less drift you’ll get across the sequence.
Pro tip ✅
Add every visual rule you care about directly into the prompt text: color palette, background color, illustration style, line weight, character clothing. Midjourney V7 reads these details — use them as a contract with the model.
This is where the actual consistency magic lives. Midjourney V7 supports two reference parameters that do different jobs and you should understand both before using either.
--cref [image URL] references a character’s appearance — face, body proportions, clothing — and tries to maintain it across generations. It’s imperfect with non-photorealistic styles, but it significantly reduces drift compared to prompting alone. --sref [image URL] locks the visual style — colors, texture, rendering approach — without tying you to a specific character. For carousel sequences, you’ll often use both together.
Upload your anchor image to Discord (where Midjourney runs), copy its URL, and use it in your next prompt like this:
flat design infographic illustration, woman in coral blazer pointing at bar chart, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, slide 2 of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [YOUR_ANCHOR_IMAGE_URL] --sref [YOUR_ANCHOR_IMAGE_URL]Run this for slide 2. Compare it against slide 1. If the style holds, you’ve found your base formula — keep it and run through slides 3 to 10 by changing only the content description (what the character is doing or what visual element is shown).
Warning ⚠️
The –cref parameter works best with photorealistic or semi-realistic images. With flat illustration styles, character consistency will still drift — expect slight variations in facial features or proportions. This is a known limitation, not a bug in your workflow.
Writing a fresh prompt from scratch for each slide is how you guarantee visual drift. Instead, create a template where only the content description changes — everything else stays identical across all 10–30 slides.
Here’s what that template structure looks like for a financial tips carousel:
flat design infographic illustration, woman in coral blazer [ACTION/CONTENT], clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, [SLIDE NUMBER] of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]For each slide, you swap out [ACTION/CONTENT] and [SLIDE NUMBER] only. Everything else — including the --cref and --sref URLs — stays locked. Here are five ready-to-adapt variants for a financial tips series:
flat design infographic illustration, woman in coral blazer holding piggy bank and smiling, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, slide 3 of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]flat design infographic illustration, woman in coral blazer sitting at desk with laptop showing graphs, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, slide 4 of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]flat design infographic illustration, woman in coral blazer holding magnifying glass over document, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, slide 5 of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]flat design infographic illustration, woman in coral blazer presenting large calendar with highlighted dates, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, slide 6 of financial tips series, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]flat design infographic illustration, woman in coral blazer holding trophy with coin symbols, clean white background, minimal geometric shapes, coral and navy color palette, consistent line weight, editorial style, final slide of financial tips series, coral and navy color palette, celebration mood, 9:16 aspect ratio --ar 9:16 --v 7 --style raw --cref [URL] --sref [URL]
Pro tip ✅
Keep a plain text file with your locked template. Before each generation session, paste it, swap the one variable, and fire. This prevents typos from breaking consistency and keeps your session fast — you’re not retyping 15 parameters every time.
Generate each slide one at a time. Midjourney gives you four variations per prompt — upscale the one that best matches your anchor image, not necessarily the most visually impressive one in isolation. Carousel consistency wins over individual slide brilliance every time.
After generating all slides, do a comparison pass. Line up your thumbnails in a folder and look at them together. Reject any slide where the color palette drifted, the character looks significantly different, or the background changed unexpectedly. Regenerate those with the same prompt — sometimes a second pass gives you a better result, and sometimes you need to tweak one word to steer Midjourney back on track.
Pro tip ✅
Name your files sequentially as you upscale them:
carousel_01.png,carousel_02.png, and so on. Midjourney’s download filenames are chaos — job IDs and timestamps. Rename immediately or you’ll spend 20 minutes sorting them before upload.
Midjourney V7 is still unreliable at rendering legible text. For carousel posts that need headline text or tip numbers on each slide, generate your images clean and add text in Canva, Adobe Express, or Figma. Export the final composited versions — these are what you upload to your scheduling platform, not the raw Midjourney outputs.
This step also gives you a quality control checkpoint. You’ll immediately spot which images don’t work compositionally for the text overlay you need, and you can regenerate those before you’re deep in a scheduling workflow.
Avoid 🚫
Don’t ask Midjourney to generate text as part of your carousel slides. Even in V7, text rendering produces garbled, decorative-looking nonsense that requires manual fixing. Use Midjourney for visuals, use design tools for type — that’s the correct division of labor.
Neither Later nor Buffer has a direct Midjourney integration. You’re downloading images from Discord, organizing them locally, and uploading manually — that’s the actual pipeline in 2026. Both platforms handle it fine; it’s just not automated.
In Later, create a new post, select Instagram Carousel as the post type, and drag your numbered images in sequence. Later lets you reorder before publishing, so if you missequence something it’s fixable. Add your captions per slide in the panel on the right. Later’s visual content calendar makes it easy to see your carousel as a unit before scheduling.
In Buffer, the carousel post builder works similarly — upload multiple images, set the order, write your caption (Buffer uses a single caption for the whole carousel on Instagram, unlike Later which supports per-slide captions on some plans). Schedule your publish time and you’re done.
Pro tip ✅
If you want to approximate automation, Later’s API and Buffer’s API both support programmatic post creation with image URLs. A technical user can write a script that takes a folder of numbered images, uploads them to the platform’s media endpoint, creates a scheduled carousel post, and logs the result. It’s not trivial, but it’s the closest thing to a real pipeline available right now without third-party middleware.
Once you’ve generated one carousel with a locked visual style, you’ve built an asset: your anchor image and your template prompt. The next carousel in the same series takes a fraction of the time because your style reference is already established.
Save your anchor images in a dedicated folder organized by series. When you’re starting a new carousel for the same brand or account, pull the anchor, paste it into --cref and --sref, update the topic-specific content descriptions, and you’re generating slide 2 within minutes of starting. That’s the real efficiency gain — not a batch mode, but a disciplined system you get faster at every time you use it.
Pro tip ✅
For accounts posting multiple carousel series, create a style reference library: one anchor image per series, named clearly, stored somewhere permanent. Midjourney’s Discord history is not a permanent archive — images expire. Download and store every anchor image you plan to reuse.
Generate anchor image with full detailed prompt. Upscale and save. Build prompt template with --cref and --sref pointing to anchor. Generate slides by swapping content description only. Upscale and rename sequentially. Add text overlays in Canva or Figma. Upload to Later or Buffer in order. Schedule and repeat with the same anchor next time.
Ten slides at four variations each means roughly 40 Midjourney generations to pick from — fast GPU hours on Standard or Pro plan. Budget accordingly, especially if you’re running multiple carousels per week. At $30/month for Standard, you get 15 hours of fast GPU time — enough for a consistent weekly carousel workflow with room to spare.
The gap between what AI carousel tutorials promise and what Midjourney actually delivers is wide — but the real workflow is solid once you stop expecting a magic button. The --cref and --sref parameters in V7 give you genuine style consistency that would have taken a professional illustrator hours to replicate. The manual steps — prompt templating, sequential naming, design tool text overlays, platform upload — take maybe 90 minutes for a ten-slide carousel once you’ve done it twice.
No tool in 2026 automates this end-to-end without custom engineering. But with a good anchor image and a disciplined template, you can produce carousel content that looks like it came from a design agency — consistently, repeatably, and on a creator budget. That’s the actual offer on the table. It’s a good one.
