What Instagram, LinkedIn, and X Do to Oversized Images

Instagram re-encodes a heavy JPEG the second you hit Share. LinkedIn and X do the same once you cross their size caps. You lose sharpness, color fidelity, and metadata the moment the file hits their pipeline. Compress to their limits yourself, or each platform chooses the quality for you.

I watch drop-off everywhere. On social, the drop-off is visual: the clean export you approved in Lightroom becomes the soft, blotchy feed version your audience actually sees. That gap is not random. It is policy plus server-side compression, applied without asking you.

This post maps image file size limits on Instagram, LinkedIn, and X, what each platform does when you ignore those limits, and why pre-compressing with a tool you control (I use the pipeline at Atlas Photo for client work) beats hoping the platform is gentle.

What happens when your image exceeds a platform size limit?

Platform size limits are hard caps plus soft quality rules. A hard cap rejects the upload or forces a client-side resize. Soft rules re-encode accepted files so storage and bandwidth stay cheap. Most creators only notice the second category, because the post still goes live. The pixels do not.

When a file is too large, platforms typically do one or more of the following:

• Recompress JPEGs at a lower quality factor, often with different chroma subsampling than your export.
• Downscale long edges to a max width (commonly around 1080px on Instagram feed, higher on some LinkedIn and X surfaces).
• Strip or rewrite metadata (EXIF, IPTC, GPS, camera make/model, copyright strings).
• Convert formats (for example PNG to JPEG on X in many cases, or HEIC to JPEG on mobile upload paths).
• Strip progressive structure or color profiles and re-apply a generic sRGB assumption.

The platform is not your colorist. Once the file leaves your device, every byte is fair game for storage optimization.

If you want a single operating rule: treat Instagram, LinkedIn, and X as hostile re-encoders. Export for their limits first. Upload second.

What are Instagram image file size limits and what does Instagram do to uploads?

Instagram image file size limits sit near 30 MB per photo for many feed and story paths, with practical quality targets far below that. Feed imagery is still optimized around roughly 1080px on the short side for display. A 24 MP camera JPEG at high quality can clear the 30 MB gate and still look soft after upload, because acceptance and quality are different steps.

What Instagram actually does to your file:

1. Accepts supported formats (JPEG and PNG are the workhorses; HEIC often converts on the way in from mobile).
2. Resizes toward feed-friendly dimensions when you send excess resolution.
3. Re-encodes with aggressive JPEG settings tuned for mobile bandwidth, not print fidelity.
4. Strips most EXIF and location data on the public asset path. Copyright fields you carefully wrote often vanish from the delivered image.
5. Applies further compression for different surfaces (feed, story, profile grid thumbnails), so one master upload is not one quality.

Practical result: a 12–18 MB studio JPEG often looks worse than a 400–900 KB export that was already sized to ~1080–1440px and compressed with a controlled quality target. The large file did not protect you. It invited a harsher second encode.

I still see teams export “maximum quality” from Capture One, then blame Instagram for “ruining” skin tones. Instagram did what a CDN does under load. The fix is upstream.

For official product surface notes and changing product rules, start from Instagram Help Center and verify the current post type you use (feed, Reel cover, story, carousel). Specs drift by product surface.

What are LinkedIn image limits and how hard does LinkedIn compress?

LinkedIn image uploads for standard feed posts commonly land around an 8 MB ceiling, with recommended share dimensions near 1200×627 for link-style cards and larger square or portrait crops for native image posts. Exact caps shift by surface (personal feed, company page, ads), but the compression pattern is consistent: LinkedIn prioritizes crisp-enough UI on laptop and mobile, not gallery-grade files.

What LinkedIn does past the threshold or under the soft quality pipeline:

• Rejects or forces a client resize when you exceed the hard MB limit.
• Downscales wide graphics that exceed its display widths.
• Recompresses JPEGs so multi-MB screenshots and camera files become leaner assets.
• Often removes rich metadata from the delivered asset, which matters if you rely on embedded copyright or caption fields for proof of origin.
• Treats PNG UI graphics better for sharp edges than high-ISO photos, then still may re-encode depending on size.

LinkedIn is slightly kinder to text-in-image graphics than Instagram is to portraits, but “kinder” is not “lossless.” A 7.9 MB full-bleed campaign stillboard can still smear thin type if you leave the platform to pick quality. Pre-flatten text at the target width, then compress until file size sits well under 8 MB with readable type at 100% zoom on a phone.

Company pages and ads add another layer: ad libraries keep derivatives. If the first upload is over-compressed by LinkedIn, every placement inherits that look. Control the first file.

What are X (Twitter) image size limits and what does X strip from your file?

X image limits for static photos are commonly 5 MB per image, with multi-image posts applying the cap per frame. GIFs sit higher (often cited near 15 MB), and video is a separate pipeline. Max dimensions historically sit near 4096×4096, but the quality cliff appears long before you hit either wall.

What X does to oversized or merely inconvenient files:

• Blocks or resizes uploads over the MB cap depending on client (web vs app).
• Converts many PNGs to JPEG when the PNG is photo-like or large, which kills alpha and can introduce banding on gradients.
• Recompresses JPEGs with visible artifacts on fine hair, fabric weave, and night skies.
• Strips EXIF and most creator metadata from the public CDN URL.
• Serves different derivatives for timeline cards vs expand view, so the “full image” click is still not your original master.

If your brand relies on a PNG logo with transparency in a photo collage, test the upload. Transparency may disappear. If your legal team wants camera serials or copyright IPTC left intact for dispute trails, assume X will not preserve them on the public file. Keep the master offline; post a derivative built for the feed.

X Help documents current media rules here: X Help Center media. Treat any number you read as “until next product change,” and re-check before a major campaign.

Instagram vs LinkedIn vs X: size limits and damage patterns

Use this as a working cheat sheet for operators, not a forever spec sheet. Always re-verify before a launch week.

Platform	Typical hard size cue	What usually happens if you ignore it	Metadata fate	Format risk
Instagram	~30 MB photo; display ~1080px class	Strong re-encode + multi-surface derivatives	Most EXIF/location stripped	HEIC often becomes JPEG; heavy JPEG recompress
LinkedIn	~8 MB feed image	Resize + recompress; text can soften	Rich metadata often gone on delivery	PNG OK for UI; photos still re-encoded
X	~5 MB static image	Cap enforcement + aggressive recompress	EXIF largely stripped	PNG may become JPEG; alpha at risk

Three numbers to remember when you are mid-export: 30 / 8 / 5. Those are the rough MB fences for Instagram, LinkedIn, and X. Working well under them with pre-sized dimensions almost always beats maxing them out.

Why should you compress before upload instead of letting the platform decide?

Platform compression is a black box. Your export settings are a lab. That is the whole argument.

When you compress first:

1. You pick the quality floor. A controlled 75–85 JPEG quality at the right pixel width often beats an uncontrolled server pass on a 15 MB master.
2. You keep a known before/after. Side-by-side the master and the social derivative. Platforms will not show you their intermediate.
3. You reduce double compression. Uploading a huge high-quality JPEG that gets re-encoded is classic generation loss. Start closer to the delivery size.
4. You protect brand color within reality. Soft-proof to sRGB, embed or convert intentionally, then accept that social will still approximate. Do not leave color and chroma decisions entirely to a mobile CDN.
5. You ship faster on bad networks. Smaller, intentional files fail less on flaky conference Wi-Fi when your social manager is posting live.

I am running a tiny mental A/B every time I open a feed: does this look like a photographer exported it, or like a platform rescued storage? You can usually tell within a second on skin and sky.

A sane pre-upload checklist for all three platforms:

• Convert to sRGB.
• Resize long edge to the surface you care about (often 1080–2048px depending on platform and crop).
• Export JPEG with a visible quality check at 100% on a phone-sized preview, not only on a calibrated desktop.
• Land under the hard MB cap with headroom (for X, think well under 5 MB; for LinkedIn, under 8 MB; for Instagram, quality targets matter more than filling 30 MB).
• Strip GPS if privacy matters, and keep copyright in your own archive even if the platform discards EXIF.
• Name files for humans (brand-product-feed-1080.jpg), not for cameras (IMG_88421.JPG).

If you want that pipeline without reinventing export presets every week, run images through a dedicated compression workflow before you open the native apps. That is the product job behind Atlas Photo: get file size and visual quality under your control so Instagram, LinkedIn, and X inherit a good decision instead of inventing a worse one.

How do you pick target file sizes that still look sharp?

Target file size is a proxy for bitrate across a known pixel count. For a 1080×1350 Instagram portrait, many brand feeds look clean in the 200–600 KB band when the export is well tuned. Busy detail (foliage, fabric, city night) may need 600 KB–1.5 MB. Going to multi-megabyte files at the same dimensions rarely buys proportional sharpness on-platform, because the second encode flattens the gain.

For LinkedIn document-style graphics with type, prioritize edge sharpness over tiny files. A 300–800 KB PNG or carefully compressed JPEG at exact layout size often beats a 5 MB photo export of a slide (see PNG vs carefully compressed JPEG).

For X, stay especially lean. Timeline cards are small. A razor-sharp 400–900 KB JPEG at sensible dimensions will usually out-perform a near-cap 4.8 MB dump that X then panics-compresses.

Rule of thumb I give clients: if the file is more than about 2× your proven good export for that surface, you are paying for bytes the platform will throw away.

What metadata do you lose, and does it matter for marketing?

Metadata loss is Y for most social CDNs: metadata loss is the removal of embedded fields (camera settings, GPS, copyright, captions) from the public image file. Marketing teams feel this when:

• Legal wanted IPTC copyright on every public asset.
• You hoped GPS would help local SEO (it will not, on-platform).
• You reverse-image-search a stolen post and find no camera serial trail in the hotlink file.

Assume public social files are metadata-light. Store masters and rights data in your DAM. Publish derivatives built for the network. That split is boring and it prevents ugly surprises in brand disputes.

FAQ

Does a larger upload always look better on Instagram?

No. A larger upload often looks worse after Instagram re-encodes it. Matched pixel dimensions with a controlled quality export usually beat a multi-megabyte “max quality” master that invites a harsher second pass.

Will LinkedIn or X keep my PNG transparency and EXIF?

Do not count on either. X may convert photo-like PNGs toward JPEG and strip EXIF. LinkedIn delivers compressed derivatives that drop most embedded fields. Keep transparency-critical and rights-critical data in your own archive and test the live URL.

What is the fastest way to stop platform compression from wrecking a campaign?

Build one export preset per platform, sit under the hard MB limits with headroom, and visually QA on a phone before scheduling. Use a compression tool you control so the first public file is already the version you approved.

---

Open your last three social posts on a phone, download or screenshot the live asset, and compare it to the file still on disk. Note where edges went soft and where color flattened. Then set pixel width and file size targets for Instagram, LinkedIn, and X, and run the next batch through a pre-upload compression pass before any of those apps get a vote.

---

Find any asset in seconds. Atlas Photo is digital asset management for creative and brand teams, with early-access founder pricing for the first users. Get early access